Skip to content

An agent organization structured as a company — 15+ departments, 125+ skills, each independently installable, citing the standards and regulators that settle the question. Runs in Claude Code and ChatGPT.

Complete Never coverage across the skill library

Claudeaa45383819f8
diff --git a/plugins/corporate-strategy/skills/portfolio-strategy/SKILL.md b/plugins/corporate-strategy/skills/portfolio-strategy/SKILL.md
index ee1ca8c..2726a74 100644
--- a/plugins/corporate-strategy/skills/portfolio-strategy/SKILL.md
+++ b/plugins/corporate-strategy/skills/portfolio-strategy/SKILL.md
@@ -54,6 +54,13 @@ allocation you already had.
Force a ranking. Tiers are how everything stays funded.
+## Never
+
+- Fund a line because it is large. Fund it on marginal return.
+- Keep a line alive on sunk cost.
+- Starve a line without deciding to exit it. Slow starvation costs more than a clean exit.
+- Review the portfolio only when a line is already in trouble.
+
## Return contract
Each line with its posture and evidence, the recommended allocation and what changed from last
diff --git a/plugins/corporate-strategy/skills/strategic-alliances/SKILL.md b/plugins/corporate-strategy/skills/strategic-alliances/SKILL.md
index 329a640..5a74f5d 100644
--- a/plugins/corporate-strategy/skills/strategic-alliances/SKILL.md
+++ b/plugins/corporate-strategy/skills/strategic-alliances/SKILL.md
@@ -59,3 +59,10 @@ strategy moved.
Say it plainly and early. Partnerships die quietly for a year before anyone admits it, and that year
is the cost.
+
+## Never
+
+- Sign a partnership without naming what each side is measured on.
+- Announce before the joint work has a named owner on both sides.
+- Let a partnership run past its review date on goodwill.
+- Grant exclusivity without a performance floor that ends it.
diff --git a/plugins/customer-experience/skills/escalation-management/SKILL.md b/plugins/customer-experience/skills/escalation-management/SKILL.md
index 0a45b64..32ea827 100644
--- a/plugins/customer-experience/skills/escalation-management/SKILL.md
+++ b/plugins/customer-experience/skills/escalation-management/SKILL.md
@@ -57,3 +57,10 @@ correctly, did we communicate on time, and what would have prevented it.
Escalation volume is a health metric for the whole function. Rising escalations mean the normal path
is failing more often, and that is the thing to fix.
+
+## Never
+
+- Escalate without naming an owner. An escalation with no name on it is a broadcast.
+- Commit to a fix date on the customer's call before engineering has said one exists.
+- Let the executive sponsor become the case owner. Sponsors unblock; they do not run the case.
+- Close on the technical fix. It closes when the customer says it is closed.
diff --git a/plugins/customer-experience/skills/self-service-and-knowledge/SKILL.md b/plugins/customer-experience/skills/self-service-and-knowledge/SKILL.md
index f4f0cdd..3e3d742 100644
--- a/plugins/customer-experience/skills/self-service-and-knowledge/SKILL.md
+++ b/plugins/customer-experience/skills/self-service-and-knowledge/SKILL.md
@@ -62,3 +62,10 @@ volume for a topic falls after content ships.
Watch articles with high views *and* a high subsequent contact rate. Those are articles that are
failing to answer, and they look like your best-performing content.
+
+## Never
+
+- Write an article for a problem the product should not have. Fix the product and delete the article.
+- Publish without an owner and a review date. Stale help is worse than no help.
+- Measure a knowledge base by article count.
+- Hide the path to a human. Deflection that traps people costs more than the ticket would have.
diff --git a/plugins/customer-experience/skills/support-operations/SKILL.md b/plugins/customer-experience/skills/support-operations/SKILL.md
index 947aafc..9973964 100644
--- a/plugins/customer-experience/skills/support-operations/SKILL.md
+++ b/plugins/customer-experience/skills/support-operations/SKILL.md
@@ -68,3 +68,10 @@ by a wide margin and then blames the team.
Review a sample of resolved contacts against a rubric agreed with the team, and coach against it.
Reviewing only escalations trains for defense rather than quality.
+
+## Never
+
+- Staff to average volume. Support arrives in peaks.
+- Publish a service level you have not staffed to meet.
+- Manage on handle time. It optimizes for closing tickets, not for solving problems.
+- Let a repeat driver stay a support problem. Route it to whoever owns the cause.
diff --git a/plugins/data-analytics/skills/business-intelligence/SKILL.md b/plugins/data-analytics/skills/business-intelligence/SKILL.md
index 47aebcc..8fc6e50 100644
--- a/plugins/data-analytics/skills/business-intelligence/SKILL.md
+++ b/plugins/data-analytics/skills/business-intelligence/SKILL.md
@@ -63,3 +63,10 @@ check every figure by hand.
Dashboards accumulate. Review usage periodically and retire what nobody opens — with a notice period,
since the one person using it may be using it for something important.
+
+## Never
+
+- Build a dashboard nobody has a decision for. Start from the decision.
+- Ship a metric with two definitions live at the same time.
+- Leave a dashboard published with no owner. Unowned dashboards get trusted, then get wrong.
+- Show a number without the denominator and the window it covers.
diff --git a/plugins/executive/skills/agent-hierarchy/SKILL.md b/plugins/executive/skills/agent-hierarchy/SKILL.md
index 561eaed..8b32306 100644
--- a/plugins/executive/skills/agent-hierarchy/SKILL.md
+++ b/plugins/executive/skills/agent-hierarchy/SKILL.md
@@ -61,6 +61,13 @@ decision is raised, assign it the next number immediately — before it is answe
lettered options with an explicit recommendation. Never renumber, never reuse a number, and record
resolutions in place rather than deleting them.
+## Never
+
+- Add an agent before the surface it will own exists. The map comes first, the agent second.
+- Let two agents hold a write claim on the same path. Overlap resolves itself as a race.
+- Hand-edit a generated artifact instead of the source it derives from.
+- Skip the guard because the change is small. Small changes are how surfaces drift.
+
## References
- `references/playbook.md` — the full 415-line playbook: surface splitting, the guard, the
diff --git a/plugins/executive/skills/business-growth-consultant/SKILL.md b/plugins/executive/skills/business-growth-consultant/SKILL.md
index c76e528..c9f3cc3 100644
--- a/plugins/executive/skills/business-growth-consultant/SKILL.md
+++ b/plugins/executive/skills/business-growth-consultant/SKILL.md
@@ -68,6 +68,13 @@ Watch for metrics that improve while the business worsens: revenue up with margi
with activation down, traffic up with conversion down. Each means the mix shifted, and the headline
number is hiding it.
+## Never
+
+- Recommend tactics before naming the constraint. Effort applied away from the constraint does nothing.
+- Stack initiatives until none of them gets the attention it needs to work.
+- Present a plan with no way to tell within a quarter whether it is working.
+- Recommend more spend into a funnel that does not convert what it already receives.
+
## Return contract
1. **The constraint**, named, with the evidence.
diff --git a/plugins/executive/skills/ceo-advisor/SKILL.md b/plugins/executive/skills/ceo-advisor/SKILL.md
index 0582c0e..37c7df6 100644
--- a/plugins/executive/skills/ceo-advisor/SKILL.md
+++ b/plugins/executive/skills/ceo-advisor/SKILL.md
@@ -65,6 +65,13 @@ Where a commitment was missed repeatedly, the interesting question is not why th
whether the commitment was ever realistic, or whether it is not actually a priority — either answer
is useful, and both are better than a fourth attempt.
+## Never
+
+- Soften an assessment because it is unwelcome. That is the one thing this role exists for.
+- Give advice without saying what would change your mind.
+- Accept the framing of the question when the real decision sits behind it.
+- Lay out options and stop. Say which one you would pick.
+
## Return contract
1. **The real decision**, restated.
diff --git a/plugins/it-operations/skills/virtualization-operations/SKILL.md b/plugins/it-operations/skills/virtualization-operations/SKILL.md
index b22a05f..6f68110 100644
--- a/plugins/it-operations/skills/virtualization-operations/SKILL.md
+++ b/plugins/it-operations/skills/virtualization-operations/SKILL.md
@@ -80,3 +80,10 @@ for.
Hybrid is the normal end state, not a transitional embarrassment. What makes it painful is running
two operating models with two sets of habits, so decide deliberately which one owns identity,
monitoring, and backup rather than letting each side answer differently.
+
+## Never
+
+- Provision a guest without an owner, a lifetime, and a decommission trigger.
+- Size a cluster with no headroom to lose a node during maintenance.
+- Grow a guest's resources before checking whether the guest itself is misbehaving. Oversizing hides the bug and eats the consolidation ratio.
+- Move a workload to cloud to escape a capacity problem you have not diagnosed on-premises.
diff --git a/plugins/legal-risk/skills/corporate-governance/SKILL.md b/plugins/legal-risk/skills/corporate-governance/SKILL.md
index 64f4f7a..db76664 100644
--- a/plugins/legal-risk/skills/corporate-governance/SKILL.md
+++ b/plugins/legal-risk/skills/corporate-governance/SKILL.md
@@ -63,3 +63,10 @@ destroy anything under a legal hold, and know who can place one.
Some responsibilities sit between functions and get dropped. When one appears, do not debate the
right home in the abstract — assign it to whoever bears the consequence if it fails, and record the
assignment. An owner who is imperfect beats an owner who is undecided.
+
+## Never
+
+- Take an action that requires board or member approval and paper it afterward.
+- Let the corporate record live in someone's inbox.
+- Sign outside a documented delegation of authority because the deal is urgent.
+- Destroy records on schedule while a legal hold is in place. A hold suspends the schedule.
diff --git a/plugins/legal-risk/skills/enterprise-risk/SKILL.md b/plugins/legal-risk/skills/enterprise-risk/SKILL.md
index 7a2587e..8237f72 100644
--- a/plugins/legal-risk/skills/enterprise-risk/SKILL.md
+++ b/plugins/legal-risk/skills/enterprise-risk/SKILL.md
@@ -53,3 +53,10 @@ found then cannot be fixed retroactively.
Leadership needs the few risks whose residual exposure is above appetite, what is being done, and
what needs a decision. Not the whole register. A risk report that requires reading forty rows to
find the three that matter will not be read.
+
+## Never
+
+- Score residual risk on controls that are planned rather than operating.
+- Accept a risk without naming who accepted it and when it is reviewed again.
+- Keep a register with no review cadence. That is documentation, not risk management.
+- Close a risk because the project that raised it ended.
diff --git a/plugins/legal-risk/skills/privacy-and-data-protection/SKILL.md b/plugins/legal-risk/skills/privacy-and-data-protection/SKILL.md
index 86f9eb9..45e9551 100644
--- a/plugins/legal-risk/skills/privacy-and-data-protection/SKILL.md
+++ b/plugins/legal-risk/skills/privacy-and-data-protection/SKILL.md
@@ -55,3 +55,10 @@ data is located across systems, and the deadline. Locating the data is the part
For breaches, know your notification clock before you need it — several regimes measure it in hours
from awareness. Decide in advance who determines that awareness has occurred.
+
+## Never
+
+- Collect data because it may be useful later. Purpose first, then collection.
+- Retain personal data past the period you published.
+- Send personal data to a vendor before the contract terms and the transfer basis are in place.
+- Load production personal data into a test environment.
diff --git a/plugins/operations/skills/process-design/SKILL.md b/plugins/operations/skills/process-design/SKILL.md
index 997f222..d2fdd2c 100644
--- a/plugins/operations/skills/process-design/SKILL.md
+++ b/plugins/operations/skills/process-design/SKILL.md
@@ -51,6 +51,13 @@ system: required fields, blocking gates, defaults that are correct, automated ha
**Never fix a recurring failure with a reminder.** If people are forgetting a step, the system
permits forgetting it. Fix that.
+## Never
+
+- Redesign from the documented process. Map what people actually do first.
+- Optimize a step that is not the constraint. Everything you speed up ahead of it just queues.
+- Add an approval to prevent a failure that happened once.
+- Hand over a process with no owner and no measure. It reverts within a quarter.
+
## Return contract
Current state with elapsed and touch time, the constraint and its evidence, the redesign, what it
diff --git a/plugins/operations/skills/vendor-management/SKILL.md b/plugins/operations/skills/vendor-management/SKILL.md
index 0cf25ac..6b9b7c0 100644
--- a/plugins/operations/skills/vendor-management/SKILL.md
+++ b/plugins/operations/skills/vendor-management/SKILL.md
@@ -50,3 +50,10 @@ triples its price. For each, know the exit path and roughly what it costs — an
never been thought through is not an option, it is a hope.
Maintain your own copy of your data continuously where the vendor holds anything critical.
+
+## Never
+
+- Sign before you know what leaving costs — data export, notice period, transition support.
+- Let the vendor write the requirements you evaluate them against.
+- Reach a renewal date without having started the renewal. The auto-renew clause is their leverage.
+- Concentrate a critical dependency on one vendor without saying so out loud and pricing the risk.
diff --git a/plugins/product/skills/brand-identity/SKILL.md b/plugins/product/skills/brand-identity/SKILL.md
index 0547fcb..09f90bf 100644
--- a/plugins/product/skills/brand-identity/SKILL.md
+++ b/plugins/product/skills/brand-identity/SKILL.md
@@ -44,3 +44,10 @@ Collect real surfaces — product, marketing site, email, social, sales collater
for: palette drift, more type families than the system defines, logo variants nobody sanctioned, and
tone that changes between surfaces. Report by severity, and separate what breaks the system from
what merely was not specified.
+
+## Never
+
+- Pick color and type before you know what the brand has to sound like and what it will sit next to.
+- Ship a palette you have not checked for contrast at real sizes on real backgrounds.
+- Use color alone to carry state, meaning, or error.
+- Add a typeface because a page needs variety. Variety comes from weight, size, and spacing inside the system you already have.
diff --git a/plugins/product/skills/design-styles/SKILL.md b/plugins/product/skills/design-styles/SKILL.md
index 596b302..6e647b5 100644
--- a/plugins/product/skills/design-styles/SKILL.md
+++ b/plugins/product/skills/design-styles/SKILL.md
@@ -51,3 +51,10 @@ Fails when: applied to high-frequency tools, where the motion and padding become
3. Apply completely. A utilitarian grid with soft shadows is not a hybrid.
4. Do not change layout structure to suit the style. If structure must change, that is a separate
decision, stated as one.
+
+## Never
+
+- Mix two styles on one surface because each has a part you like. Pick one and hold it.
+- Choose a style before you know the content it has to carry and who is reading it.
+- Treat a style as decoration applied at the end. It decides density, spacing, and hierarchy from the first screen.
+- Copy a reference into a layout your real content cannot fill. Empty space that was supposed to hold something reads as broken.
diff --git a/plugins/product/skills/design-system/SKILL.md b/plugins/product/skills/design-system/SKILL.md
index 7f81431..3d087e6 100644
--- a/plugins/product/skills/design-system/SKILL.md
+++ b/plugins/product/skills/design-system/SKILL.md
@@ -41,6 +41,13 @@ a polish item; it is whether people can use it.
- Never remove a token or component because it looks unused. You cannot see every consumer from
inside the system. Deprecate, announce, then remove.
+## Never
+
+- Add a component for a single screen. A one-use component is a snippet, not a system.
+- Hard-code a value a token already covers. The exception becomes the next inconsistency.
+- Redefine what an existing token means instead of adding one. Changing a semantic silently redecorates everything downstream.
+- Ship a component without its states documented — empty, loading, error, disabled.
+
## Return contract
Report tokens added or changed, components affected, anything now inconsistent with the system, and
diff --git a/plugins/product/skills/interface-redesign/SKILL.md b/plugins/product/skills/interface-redesign/SKILL.md
index 3dfdcf3..4b629f4 100644
--- a/plugins/product/skills/interface-redesign/SKILL.md
+++ b/plugins/product/skills/interface-redesign/SKILL.md
@@ -47,6 +47,13 @@ Fix in this order, because each makes the next easier to see:
Rebuild only when the structure is wrong — when the information architecture does not match how
people work. Everything else is a restyle, and a restyle that ships beats a rebuild that stalls.
+## Never
+
+- Restyle a screen whose problem is structural. New paint on a broken flow buys nothing.
+- Change navigation and visual language in the same release. When something breaks you will not know which one did it.
+- Remove a path because it looks unused without checking who uses it and what they do next.
+- Ship a redesign with no way back. Keep the old path reachable until the new one is proven.
+
## Return contract
Findings by category with severity, the sequenced plan, what you would do first if only one day were
diff --git a/plugins/product/skills/ux-product-auditor/SKILL.md b/plugins/product/skills/ux-product-auditor/SKILL.md
index 5486a8a..2bb193c 100644
--- a/plugins/product/skills/ux-product-auditor/SKILL.md
+++ b/plugins/product/skills/ux-product-auditor/SKILL.md
@@ -86,3 +86,10 @@ Where the right fix is uncertain, propose the cheapest way to find out rather th
Rank by users affected × severity ÷ effort. Then state the one thing to fix first, and be willing to
say that most of the list is not worth doing yet. A prioritized audit is more useful than a complete
one.
+
+## Never
+
+- Report a finding without the path that produces it. A claim nobody can reproduce gets ignored.
+- Rank findings by how easy they are to fix. Rank by cost to the user, then note which ones are cheap.
+- Audit a staging build or a marketing page that real users never touch.
+- Call a preference a defect. Say what breaks and for whom, or leave it out.
diff --git a/plugins/product/skills/visual-reference-generation/SKILL.md b/plugins/product/skills/visual-reference-generation/SKILL.md
index 95d2fc9..8893f2b 100644
--- a/plugins/product/skills/visual-reference-generation/SKILL.md
+++ b/plugins/product/skills/visual-reference-generation/SKILL.md
@@ -41,3 +41,10 @@ not the screenshot.
Say explicitly what in the reference is **direction** and what is **placeholder**. A developer
handed a concept will otherwise implement the lorem ipsum faithfully.
+
+## Never
+
+- Generate before the content, states, and constraints are settled. A reference built on placeholder copy hides every real problem.
+- Present a generated image as a working design. Say what it is.
+- Generate with filler text where real string lengths decide the layout.
+- Hand off an image as the spec. Hand off the tokens, spacing, and states behind it.
diff --git a/plugins/revenue/skills/activation/SKILL.md b/plugins/revenue/skills/activation/SKILL.md
index 48df8c1..bc1b2b7 100644
--- a/plugins/revenue/skills/activation/SKILL.md
+++ b/plugins/revenue/skills/activation/SKILL.md
@@ -56,3 +56,10 @@ for.
Percentage reaching the activation moment, and time to reach it. Signup conversion alone will
happily improve while activation falls.
+
+## Never
+
+- Pick an activation moment because it is easy to measure. It has to be the point where the user gets the value they came for.
+- Add a step to onboarding without removing one. Every screen between signup and value costs you users.
+- Ask at signup for anything you do not need to deliver the first result. Collect the rest later, in context.
+- Count signups as activation. A user who never reached the value moment did not activate.
diff --git a/plugins/revenue/skills/outbound-prospecting/SKILL.md b/plugins/revenue/skills/outbound-prospecting/SKILL.md
index 05e2801..bdae41b 100644
--- a/plugins/revenue/skills/outbound-prospecting/SKILL.md
+++ b/plugins/revenue/skills/outbound-prospecting/SKILL.md
@@ -55,3 +55,10 @@ Use a subdomain for outbound so a reputation problem cannot take down your trans
statement.
- **Replies, no meetings** — a qualification problem: you are reaching people who cannot act.
- **Meetings, no pipeline** — the segment is wrong.
+
+## Never
+
+- Send before the list is qualified. A good message to the wrong person is still a bad message.
+- Send outbound from the domain that carries your customer and billing mail. Prospecting burns sender reputation; keep it on a separate domain.
+- Keep sequencing someone who replied. Pull them out the moment a human is in the thread.
+- Judge a sequence on open rates. Opens are unreliable and unactionable; judge on replies and meetings held.
diff --git a/plugins/revenue/skills/pricing-and-packaging/SKILL.md b/plugins/revenue/skills/pricing-and-packaging/SKILL.md
index 9f4ee98..0a41ab1 100644
--- a/plugins/revenue/skills/pricing-and-packaging/SKILL.md
+++ b/plugins/revenue/skills/pricing-and-packaging/SKILL.md
@@ -52,3 +52,10 @@ buy one conversion and lose the account.
- Announce with real notice and a clear reason.
- Change one thing at a time — price and packaging together makes the result unreadable.
- Model the downside first: at what churn rate does this increase lose money?
+
+## Never
+
+- Change price without telling existing customers before it takes effect and honoring what they already signed.
+- Pick a pricing metric the customer cannot predict. If they cannot forecast the bill, they will not sign.
+- Add a tier to close a single deal. Tiers are a permanent tax on every conversation after it.
+- Discount without taking something back in exchange — term, scope, payment timing, a reference. A free discount resets the price for everyone.
diff --git a/plugins/revenue/skills/referral-programs/SKILL.md b/plugins/revenue/skills/referral-programs/SKILL.md
index 338dffa..09ccbad 100644
--- a/plugins/revenue/skills/referral-programs/SKILL.md
+++ b/plugins/revenue/skills/referral-programs/SKILL.md
@@ -53,3 +53,10 @@ you already had.
Track referred-customer retention against baseline. If referred customers retain worse, the
incentive is buying the wrong behavior and the program is losing money while appearing to work.
+
+## Never
+
+- Launch referrals before the product retains. A referral program amplifies whatever is already there, churn included.
+- Pay a reward before the referred account is real — activated, paid, and past the refund window.
+- Ask for a referral in the same breath as an apology or an open support escalation.
+- Run affiliates and customer referrals on the same terms. They attract different people and need different controls.
diff --git a/plugins/revenue/skills/revenue-operations/SKILL.md b/plugins/revenue/skills/revenue-operations/SKILL.md
index 8f2d66a..d856ba7 100644
--- a/plugins/revenue/skills/revenue-operations/SKILL.md
+++ b/plugins/revenue/skills/revenue-operations/SKILL.md
@@ -71,6 +71,13 @@ be unusable within two quarters.
Never require a field whose value is not used in a decision. Every unnecessary field trains sellers
to enter garbage in all of them.
+## Never
+
+- Change a definition without restating history on the new one. A metric that moved because you redefined it is not a result.
+- Report a forecast number you cannot trace back to named deals.
+- Let two systems each keep their own version of the same field. Pick the source of truth and make the other read from it.
+- Score leads with a model you cannot explain to the reps who have to work them.
+
## Return contract
State the definitional gaps found, the process change proposed, what it costs sellers in time, and
diff --git a/plugins/revenue/skills/sales-enablement/SKILL.md b/plugins/revenue/skills/sales-enablement/SKILL.md
index c3a2368..b8f471b 100644
--- a/plugins/revenue/skills/sales-enablement/SKILL.md
+++ b/plugins/revenue/skills/sales-enablement/SKILL.md
@@ -48,3 +48,10 @@ moment inside five minutes when time gets cut.
Collateral rots. Every asset needs an owner and a review date, and anything referencing pricing,
competitors, or product capability needs checking every quarter. A battlecard describing a
competitor's old product loses deals.
+
+## Never
+
+- Build collateral before you know where deals actually stall. Ask the reps and read the losses first.
+- Ship a deck or one-pager with no owner and no review date. Unmaintained collateral is worse than none.
+- Order a demo around the product's feature list instead of the order the buyer's problem unfolds in.
+- Count enablement as done at delivery. It is done when reps use it in live deals.
diff --git a/plugins/security/skills/access-and-identity/SKILL.md b/plugins/security/skills/access-and-identity/SKILL.md
index 8296b40..02f63ba 100644
--- a/plugins/security/skills/access-and-identity/SKILL.md
+++ b/plugins/security/skills/access-and-identity/SKILL.md
@@ -64,3 +64,10 @@ unnecessary or belongs to someone who left.
Look for: permissions granted to individuals rather than roles, roles nobody can define, standing
administrative access, accounts whose owner has left, service credentials with no owner, and systems
outside SSO. Each is a specific fix, and the list is nearly always the same list.
+
+## Never
+
+- Grant standing access where time-bound access would do the same job.
+- Leave an account active while an offboarding ticket works its way through. Cut access first, reconcile after.
+- Share a credential between people. If two people can use it, no log tells you which one did.
+- Approve your own access request, or review a group you belong to.
diff --git a/plugins/security/skills/incident-response/SKILL.md b/plugins/security/skills/incident-response/SKILL.md
index f8ec3bb..b3c143c 100644
--- a/plugins/security/skills/incident-response/SKILL.md
+++ b/plugins/security/skills/incident-response/SKILL.md
@@ -65,3 +65,10 @@ same incident recurs.
The plan matters less than having run it. Exercise once a year at minimum: a tabletop against a
realistic scenario finds the gaps — who has authority out of hours, where the credentials are, who
calls counsel — at a time when finding them is free.
+
+## Never
+
+- Rebuild or wipe a compromised host before evidence is captured.
+- Let the person running the technical response also own external communication.
+- Close an incident before you can say how entry happened and that the path is shut.
+- Speculate about cause or attribution outside the response channel while the incident is open.
diff --git a/plugins/security/skills/security-architecture-review/SKILL.md b/plugins/security/skills/security-architecture-review/SKILL.md
index ef64836..acfd52a 100644
--- a/plugins/security/skills/security-architecture-review/SKILL.md
+++ b/plugins/security/skills/security-architecture-review/SKILL.md
@@ -61,6 +61,13 @@ What data leaves, under what agreement, with what access, and what happens if th
Scope credentials to the minimum, prefer short-lived tokens, and know how to revoke without an
outage.
+## Never
+
+- Approve a design on the promise of a control with no named owner and no date.
+- Accept "it is internal" as an authorization boundary.
+- Review the diagram instead of the change. Read what is actually being built.
+- Sign off on a third-party integration without knowing what data leaves and under what terms.
+
## Return contract
Findings by severity, each with: the concrete attack, what the attacker gains, whether it blocks
diff --git a/plugins/security/skills/vulnerability-management/SKILL.md b/plugins/security/skills/vulnerability-management/SKILL.md
index 650d08b..575c48d 100644
--- a/plugins/security/skills/vulnerability-management/SKILL.md
+++ b/plugins/security/skills/vulnerability-management/SKILL.md
@@ -66,3 +66,10 @@ good-faith reporter as an adversary is how a private report becomes a public one
Leadership needs: what is exposed and unfixed past SLA, the trend, and what needs a decision. Not
the count of findings, which mostly measures how much you scanned.
+
+## Never
+
+- Rank on CVSS alone. A score is not exploitability in your environment.
+- Accept a risk without a named accepting owner and a date it gets revisited.
+- Push a patch to production without knowing what it changes and how to roll it back.
+- Report a shrinking backlog that shrank by reclassification rather than by remediation.
diff --git a/plugins/technology/skills/ai-workflow-architect/SKILL.md b/plugins/technology/skills/ai-workflow-architect/SKILL.md
index d8ce890..8208569 100644
--- a/plugins/technology/skills/ai-workflow-architect/SKILL.md
+++ b/plugins/technology/skills/ai-workflow-architect/SKILL.md
@@ -86,3 +86,10 @@ harder ones, and the first automation teaches you what the second should look li
Data leaving your control, model output reaching customers unreviewed, a silent dependency on a
vendor's pricing, and the maintenance burden landing on one person. Name the owner of each before
building, not after.
+
+## Never
+
+- Automate a process nobody has written down. You will automate the exceptions along with the work.
+- Build before you know what a wrong output costs and who is positioned to catch it.
+- Put a model where a rule would do. Deterministic steps should stay deterministic.
+- Ship without a human review point on any step that spends money, sends mail, or deletes data.
diff --git a/plugins/technology/skills/prompt-optimizer/SKILL.md b/plugins/technology/skills/prompt-optimizer/SKILL.md
index ed34c42..f6cd6eb 100644
--- a/plugins/technology/skills/prompt-optimizer/SKILL.md
+++ b/plugins/technology/skills/prompt-optimizer/SKILL.md
@@ -86,3 +86,10 @@ Do not assume a prompt transfers. Models differ in how they weight system versus
how they handle long context, and how they respond to formatting. Re-test on the target model, and
be especially suspicious of prompts tuned through many small iterations — those are often fitted to
one model's quirks.
+
+## Never
+
+- Rewrite a prompt before reading the failing outputs. You will fix the wrong thing.
+- Change more than one thing between test runs.
+- Judge a change on a single output. Sampling variance will fool you.
+- Keep an instruction because it was already there. Every line has to earn its tokens.
diff --git a/plugins/technology/skills/skill-authoring/SKILL.md b/plugins/technology/skills/skill-authoring/SKILL.md
index 3b18e80..139e922 100644
--- a/plugins/technology/skills/skill-authoring/SKILL.md
+++ b/plugins/technology/skills/skill-authoring/SKILL.md
@@ -47,3 +47,10 @@ family into one skill with references, over adding a near-neighbor.
Write three requests that should trigger it and two that should not, and check the description
actually discriminates. If a near-miss request would pull it in, tighten the description.
+
+## Never
+
+- Ship a skill whose description does not say when to use it. A skill that never triggers is dead weight.
+- Write a skill that restates what the model already does well without prompting.
+- Let two skills claim the same trigger. Move the boundary or merge them.
+- Ship without running the skill against the case that motivated writing it.
diff --git a/plugins/technology/skills/systematic-debugging/SKILL.md b/plugins/technology/skills/systematic-debugging/SKILL.md
index 3023c49..b577ec0 100644
--- a/plugins/technology/skills/systematic-debugging/SKILL.md
+++ b/plugins/technology/skills/systematic-debugging/SKILL.md
@@ -34,6 +34,13 @@ explanation has not fixed anything — it has moved the failure somewhere you ar
knowing what threw it.
- **Trusting the error message's location.** Where it surfaced is rarely where it originated.
+## Never
+
+- Close a bug without a reproduction that failed before the fix and passes after it.
+- Deploy a change to production to find out whether it works. Reproduce somewhere you can observe first.
+- Change the environment and the code in the same step.
+- Leave a debugging aid in the code as the fix — a widened timeout, a disabled check, a swallowed exception.
+
## Return contract
State the reproduction, the proven cause, the fix, and the verification that the fix addresses that
diff --git a/plugins/technology/skills/test-driven-development/SKILL.md b/plugins/technology/skills/test-driven-development/SKILL.md
index 6f0f644..da55936 100644
--- a/plugins/technology/skills/test-driven-development/SKILL.md
+++ b/plugins/technology/skills/test-driven-development/SKILL.md
@@ -40,6 +40,13 @@ claim.
- Never mock what you own — mock the network and the clock, not your own modules. Mocking your own
code tests the mock.
+## Never
+
+- Write the test after the code and call it test-driven. You will test what you built rather than what was required.
+- Skip watching the test fail. A test that has never failed proves nothing.
+- Delete or skip a failing test to unblock a release. Fix it, or revert what broke it.
+- Treat a coverage number as the goal. It counts lines executed, not behavior pinned down.
+
## Return contract
Report the tests added, what each pins down, what is deliberately untested and why, and the actual