Open-source AI penetration testing tool to find and fix your app’s vulnerabilities.
Python agentsai-hackingai-penetration-testingai-pentestingai-securityartificial-intelligencebug-bountycode-qualityctf-toolscybersecuritycybersecurity-toolsethical-hackinghackingllm-securityoffensive-securitypenetration-testingpentesting-toolsred-teamingsecuritysecurity-automation
refactor(notes): drop disk persistence + shared-wiki prose
0xallam81703e286f2f
diff --git a/HARNESS_WIKI.md b/HARNESS_WIKI.mdindex 0b876f9f..a00341f7 100644--- a/HARNESS_WIKI.md+++ b/HARNESS_WIKI.md@@ -211,7 +211,7 @@ Snapshots (`state.model_dump()`) are stored verbatim in the agent-graph node whe**Identity injection**: parent task is wrapped in `<agent_delegation>...<agent_identity>` so the child knows its name/ID and is told to never echo it (`:238-266`).-**Finish from subagent** (`agent_finish`, `:567-685`): only callable when `parent_id != None`. Builds `<agent_completion_report>` XML with summary/findings/recommendations and pushes it onto the parent's inbox. If the agent is whitebox, the wiki note is updated with a delta.+**Finish from subagent** (`agent_finish`, `:567-685`): only callable when `parent_id != None`. Builds `<agent_completion_report>` XML with summary/findings/recommendations and pushes it onto the parent's inbox.**Finish from root** (`finish_scan`, `tools/finish/finish_actions.py:86-149`): only callable from root (`parent_id is None`); blocks if any sibling/child agent is still `running`/`stopping`. Triggers `tracer.update_scan_final_fields()` which writes `penetration_test_report.md`.@@ -382,13 +382,11 @@ Supports HTTPQL filter syntax for request queries. Pagination (`offset`, `limit`Hardcoded port `48080`. Caido v0.56.0 pinned in `containers/Dockerfile` (override via `--build-arg CAIDO_VERSION=...`).-#### Notes (`strix/tools/notes/`) — `create_note`, `list_notes`, `get_note`, `update_note`, `delete_note` (+ internal `append_note_content`)-In-memory dict + JSONL persistence at `{run_dir}/notes/notes.jsonl`. Wiki-category notes additionally rendered as Markdown to `{run_dir}/wiki/{note_id}-{title}.md` so they're human-readable artifacts of the run.+#### Notes (`strix/tools/notes/`) — `create_note`, `list_notes`, `get_note`, `update_note`, `delete_note`+Pure in-memory dict, shared across every agent in the same scan for the lifetime of the process. **Not persisted** — process exit clears the lot.Categories: `general | findings | methodology | questions | plan | wiki`. IDs are 5-char UUID hex (collision-retry up to 20 attempts). Thread-safe via RLock. List preview defaults to 280 chars per note.-The wiki note in particular is **the shared whitebox knowledge base** between root and subagents; `agent_finish` for whitebox subagents auto-appends a delta (see §5.3).-#### Todos (`strix/tools/todo/`) — 6 toolsPer-agent in-memory storage; **not persisted**. Priorities `critical | high | normal | low`; statuses `pending | in_progress | done`. Bulk create/update via JSON list. IDs are 6-char UUID hex.@@ -536,8 +534,6 @@ Created and managed by `telemetry/tracer.py`. Contents:- `events.jsonl` — every span/event in append-only JSONL (thread-safe writes).- `vulnerabilities/vuln_{id}.json` — one file per finding, sorted by severity, dedup-checked.- `penetration_test_report.md` — final markdown report (executive summary + methodology + technical analysis + recommendations).-- `notes/notes.jsonl` — note ops audit log.-- `wiki/{note_id}-{slug}.md` — human-readable wiki notes.- `<target_subdir>/` — local source clones, per-target.There is **no execution checkpointing** — if the process crashes mid-run, the agent restarts from scratch on retry. Resumability is limited to the interactive-mode wait/resume on inbound messages.@@ -576,7 +572,7 @@ There **is no separate persona file**. All agents are `StrixAgent` instances usi- `parent_id` (root agent vs subagent → different finish tools, different prompts injected).- Loaded skills (root gets `root_agent`; whitebox gets `coordination/source_aware_whitebox` + `custom/source_aware_sast`).- `system_prompt_context.authorized_targets` (only set on root).-- `is_whitebox` flag (toggles wiki-note auto-update on subagent finish).+- `is_whitebox` flag (selects the whitebox skill stack).---@@ -754,7 +750,7 @@ Disabled by `STRIX_POSTHOG_TELEMETRY=0` or `STRIX_TELEMETRY=0`.| `strix/tools/terminal/` | tmux/libtmux tool. || `strix/tools/python/` | IPython tool. || `strix/tools/proxy/` | Caido GraphQL client. |-| `strix/tools/notes/` | Notes + wiki. |+| `strix/tools/notes/` | In-memory notes (shared across agents in a run). || `strix/tools/todo/` | In-memory todos. || `strix/tools/reporting/` | Vulnerability reports + CVSS. || `strix/tools/web_search/` | Perplexity. |diff --git a/strix/agents/prompts/system_prompt.jinja b/strix/agents/prompts/system_prompt.jinjaindex 6dfaf355..05673b3b 100644--- a/strix/agents/prompts/system_prompt.jinja+++ b/strix/agents/prompts/system_prompt.jinja@@ -112,11 +112,9 @@ WHITE-BOX TESTING (code provided):- MUST perform BOTH static AND dynamic analysis- Static: Use source-aware triage first to map risk quickly (`semgrep`, `ast-grep`, Tree-sitter tooling, `gitleaks`, `trufflehog`, `trivy fs`). Then review code for vulnerabilities- Static coverage floor: execute at least one structural AST mapping pass (`sg` and/or Tree-sitter) per repository and keep artifact output-- Static coverage target per repository: run one `semgrep` pass, one secrets pass (`gitleaks` and/or `trufflehog`), one `trivy fs` pass, and one AST-structural pass (`sg` and/or Tree-sitter); if any are skipped, record why in the shared wiki+- Static coverage target per repository: run one `semgrep` pass, one secrets pass (`gitleaks` and/or `trufflehog`), one `trivy fs` pass, and one AST-structural pass (`sg` and/or Tree-sitter)- Keep AST artifacts bounded and high-signal: scope to relevant paths/hypotheses, avoid whole-repo generic function dumps-- AST target selection rule: build `sg-targets.txt` from `semgrep.json` scope first (`paths.scanned`, fallback to unique `results[].path`), then run `xargs ... sg run` against that file list. Only use path-heuristic fallback if semgrep scope is unavailable, and log fallback reason in the wiki.-- Shared memory: Use notes as shared working memory; discover wiki notes with `list_notes`, then read the selected one via `get_note(note_id=...)` before analysis-- Before `agent_finish`/`finish_scan`, update the shared repo wiki with scanner summaries, key routes/sinks, and dynamic follow-up plan+- AST target selection rule: build `sg-targets.txt` from `semgrep.json` scope first (`paths.scanned`, fallback to unique `results[].path`), then run `xargs ... sg run` against that file list. Only use path-heuristic fallback if semgrep scope is unavailable.- Dynamic: Run the application and test live to validate exploitability- NEVER rely solely on static code analysis when dynamic validation is possible- Begin with fast source triage and dynamic run preparation in parallel; use static findings to prioritize live testing.diff --git a/strix/skills/coordination/source_aware_whitebox.md b/strix/skills/coordination/source_aware_whitebox.mdindex 58f0a8b6..d8849e63 100644--- a/strix/skills/coordination/source_aware_whitebox.md+++ b/strix/skills/coordination/source_aware_whitebox.md@@ -15,11 +15,10 @@ Increase white-box coverage by combining source-aware triage with dynamic valida1. Build a quick source map before deep exploitation, including at least one AST-structural pass (`sg` or `tree-sitter`) scoped to relevant paths.- For `sg` baseline, derive `sg-targets.txt` from `semgrep.json` scope first (`paths.scanned`, fallback to unique `results[].path`) and run `xargs ... sg run` on that list.- - Only fall back to path heuristics when semgrep scope is unavailable, and record the fallback reason in the repo wiki.+ - Only fall back to path heuristics when semgrep scope is unavailable.2. Run first-pass static triage to rank high-risk paths.3. Use triage outputs to prioritize dynamic PoC validation.4. Keep findings evidence-driven: no report without validation.-5. Keep shared wiki memory current so all agents can reuse context.## Source-Aware Triage Stack@@ -34,7 +33,6 @@ Coverage target per repository:- one AST structural pass (`sg` and/or `tree-sitter`)- one secrets pass (`gitleaks` and/or `trufflehog`)- one `trivy fs` pass-- if any part is skipped, log the reason in the shared wiki note## Agent Delegation Guidance@@ -42,25 +40,6 @@ Coverage target per repository:- For source-heavy subtasks, prefer creating child agents with `source_aware_sast` skill.- Use source findings to shape payloads and endpoint selection for dynamic testing.-## Wiki Note Requirement (Source Map)--When source is present, maintain one wiki note per repository and keep it current.--Operational rules:-- At task start, call `list_notes` with `category=wiki`, then read the selected wiki with `get_note(note_id=...)`.-- If no repo wiki exists, create one with `create_note` and `category=wiki`.-- Update the same wiki via `update_note`; avoid creating duplicate wiki notes for the same repo.-- Child agents should read wiki notes first via `get_note`, then extend with new evidence from their scope.-- Before calling `agent_finish`, each source-focused child agent should append a short delta update to the shared repo wiki (scanner outputs, route/sink map deltas, dynamic follow-ups).--Recommended sections:-- Architecture overview-- Entrypoints and routing-- AuthN/AuthZ model-- High-risk sinks and trust boundaries-- Static scanner summary-- Dynamic validation follow-ups-## Validation Guardrails- Static findings are hypotheses until validated.diff --git a/strix/skills/custom/source_aware_sast.md b/strix/skills/custom/source_aware_sast.mdindex f829349b..6b4f4d28 100644--- a/strix/skills/custom/source_aware_sast.md+++ b/strix/skills/custom/source_aware_sast.md@@ -15,17 +15,6 @@ Run tools from repo root and store outputs in a dedicated artifact directory:mkdir -p /workspace/.strix-source-aware```-Before scanning, check shared wiki memory:--```text-1) list_notes(category="wiki")-2) get_note(note_id=...) for the selected repo wiki before analysis-3) Reuse matching repo wiki note if present-4) create_note(category="wiki") only if missing-```--After every major source-analysis batch, update the same repo wiki note with `update_note` so other agents can reuse your latest map.-## Baseline Coverage Bundle (Recommended)Run this baseline once per repository before deep narrowing:@@ -74,8 +63,6 @@ trivy fs --scanners vuln,misconfig --timeout 30m --offline-scan \--format json --output "$ART/trivy-fs.json" . || true```-If one tool is skipped or fails, record that in the shared wiki note along with the reason.-## Semgrep First PassUse Semgrep as the default static triage pass:@@ -141,27 +128,8 @@ trivy fs --scanners vuln,misconfig --timeout 30m --offline-scan \3. Build dynamic PoCs that reproduce the suspected issue.4. Report only after dynamic validation succeeds.-## Wiki Update Template--Keep one wiki note per repository and update these sections:--```text-## Architecture-## Entrypoints-## AuthN/AuthZ-## High-Risk Sinks-## Static Findings Summary-## Dynamic Validation Follow-Ups-```--Before `agent_finish`, make one final `update_note` call to capture:-- scanner artifacts and paths-- top validated/invalidated hypotheses-- concrete dynamic follow-up tasks-## Anti-Patterns- Do not treat scanner output as final truth.- Do not spend full cycles on low-signal pattern matches.- Do not report source-only findings without validation evidence.-- Do not create multiple wiki notes for the same repository when one already exists.diff --git a/strix/skills/scan_modes/deep.md b/strix/skills/scan_modes/deep.mdindex a2687fe9..caa17994 100644--- a/strix/skills/scan_modes/deep.md+++ b/strix/skills/scan_modes/deep.md@@ -15,7 +15,6 @@ Thorough understanding before exploitation. Test every parameter, every endpoint**Whitebox (source available)**- Map every file, module, and code path in the repository-- Load and maintain shared `wiki` notes from the start (`list_notes(category="wiki")` then `get_note(note_id=...)`), then continuously update one repo note- Start with broad source-aware triage (`semgrep`, `ast-grep`, `gitleaks`, `trufflehog`, `trivy fs`) and use outputs to drive deep review- Execute at least one structural AST pass (`sg` and/or Tree-sitter) per repository and store artifacts for reuse- Keep AST artifacts bounded and query-driven (target relevant paths/sinks first; avoid whole-repo generic function dumps)@@ -31,7 +30,6 @@ Thorough understanding before exploitation. Test every parameter, every endpoint- Review file handling: upload, download, processing- Understand the deployment model and infrastructure assumptions- Check all dependency versions and repository risks against CVE/misconfiguration data-- Before final completion, update the shared repo wiki with scanner summary + dynamic follow-ups**Blackbox (no source)**- Exhaustive subdomain enumeration with multiple sources and toolsdiff --git a/strix/skills/scan_modes/quick.md b/strix/skills/scan_modes/quick.mdindex 7e8f36fe..a8824786 100644--- a/strix/skills/scan_modes/quick.md+++ b/strix/skills/scan_modes/quick.md@@ -15,7 +15,6 @@ Optimize for fast feedback on critical security issues. Skip exhaustive enumerat**Whitebox (source available)**- Focus on recent changes: git diffs, new commits, modified files—these are most likely to contain fresh bugs-- Read existing `wiki` notes first (`list_notes(category="wiki")` then `get_note(note_id=...)`) to avoid remapping from scratch- Run a fast static triage on changed files first (`semgrep`, then targeted `sg` queries)- Run at least one lightweight AST pass (`sg` or Tree-sitter) so structural mapping is not skipped- Keep AST commands tightly scoped to changed or high-risk paths; avoid broad repository-wide pattern dumps@@ -23,7 +22,6 @@ Optimize for fast feedback on critical security issues. Skip exhaustive enumerat- Identify security-sensitive patterns in changed code: auth checks, input handling, database queries, file operations- Trace user input through modified code paths- Check if security controls were modified or bypassed-- Before completion, update the shared repo wiki with what changed and what needs dynamic follow-up**Blackbox (no source)**- Map authentication and critical user flowsdiff --git a/strix/skills/scan_modes/standard.md b/strix/skills/scan_modes/standard.mdindex 13f3f707..d5f7feb5 100644--- a/strix/skills/scan_modes/standard.md+++ b/strix/skills/scan_modes/standard.md@@ -15,7 +15,6 @@ Systematic testing across the full attack surface. Understand the application be**Whitebox (source available)**- Map codebase structure: modules, entry points, routing-- Start by loading existing `wiki` notes (`list_notes(category="wiki")` then `get_note(note_id=...)`) and update one shared repo note as mapping evolves- Run `semgrep` first-pass triage to prioritize risky flows before deep manual review- Run at least one AST-structural mapping pass (`sg` and/or Tree-sitter), then use outputs for route, sink, and trust-boundary mapping- Keep AST output bounded to relevant paths and hypotheses; avoid whole-repo generic function dumps@@ -25,7 +24,6 @@ Systematic testing across the full attack surface. Understand the application be- Analyze database interactions and ORM usage- Check dependencies and repo risks with `trivy fs`, `gitleaks`, and `trufflehog`- Understand the data model and sensitive data locations-- Before completion, update the shared repo wiki with source findings summary and dynamic validation next steps**Blackbox (no source)**- Crawl application thoroughly, interact with every featurediff --git a/strix/tools/notes/tools.py b/strix/tools/notes/tools.pyindex 1818a95e..e96837da 100644--- a/strix/tools/notes/tools.py+++ b/strix/tools/notes/tools.py@@ -1,10 +1,10 @@"""Per-run notes (shared across agents).-Persisted to ``run_dir/notes/notes.jsonl`` (replayable event log) and,-for the ``wiki`` category, also rendered as Markdown to-``run_dir/wiki/<slug>.md``. Concurrent appends are serialised by a-threading.RLock so two agents writing simultaneously can't corrupt-the JSONL.+Pure in-memory state for the lifetime of one scan: the module-level+``_notes_storage`` dict is visible to every agent in the same process.+No disk I/O — process exit clears the lot. Concurrent writers are+serialised by ``_notes_lock`` since each tool entry-point dispatches+the impl onto a worker thread via ``asyncio.to_thread``."""from __future__ import annotations@@ -15,11 +15,7 @@ import loggingimport threadingimport uuidfrom datetime import UTC, datetime-from typing import TYPE_CHECKING, Any---if TYPE_CHECKING:- from pathlib import Path+from typing import Anyfrom agents import RunContextWrapper, function_tool@@ -30,162 +26,14 @@ logger = logging.getLogger(__name__)_notes_storage: dict[str, dict[str, Any]] = {}_VALID_NOTE_CATEGORIES = ["general", "findings", "methodology", "questions", "plan", "wiki"]_notes_lock = threading.RLock()-_loaded_notes_run_dir: str | None = None_DEFAULT_CONTENT_PREVIEW_CHARS = 280-def _get_run_dir() -> Path | None:- try:- from strix.telemetry.tracer import get_global_tracer-- tracer = get_global_tracer()- if not tracer:- return None- return tracer.get_run_dir()- except (ImportError, OSError, RuntimeError):- return None---def _get_notes_jsonl_path() -> Path | None:- run_dir = _get_run_dir()- if not run_dir:- return None- notes_dir = run_dir / "notes"- notes_dir.mkdir(parents=True, exist_ok=True)- return notes_dir / "notes.jsonl"---def _append_note_event(op: str, note_id: str, note: dict[str, Any] | None = None) -> None:- """Append one note operation to the run's ``notes/notes.jsonl``.-- C6: hold ``_notes_lock`` across the file open + write so two- concurrent agents can't interleave bytes mid-line.- """- notes_path = _get_notes_jsonl_path()- if not notes_path:- return- event: dict[str, Any] = {- "timestamp": datetime.now(UTC).isoformat(),- "op": op,- "note_id": note_id,- }- if note is not None:- event["note"] = note- with _notes_lock, notes_path.open("a", encoding="utf-8") as f:- f.write(f"{json.dumps(event, ensure_ascii=True)}\n")---def _load_notes_from_jsonl(notes_path: Path) -> dict[str, dict[str, Any]]:- hydrated: dict[str, dict[str, Any]] = {}- if not notes_path.exists():- return hydrated- with notes_path.open(encoding="utf-8") as f:- for raw_line in f:- line = raw_line.strip()- if not line:- continue- try:- event = json.loads(line)- except json.JSONDecodeError:- continue- op = str(event.get("op", "")).strip().lower()- note_id = str(event.get("note_id", "")).strip()- if not note_id or op not in {"create", "update", "delete"}:- continue- if op == "delete":- hydrated.pop(note_id, None)- continue- note = event.get("note")- if not isinstance(note, dict):- continue- existing = hydrated.get(note_id, {})- existing.update(note)- hydrated[note_id] = existing- return hydrated---def _ensure_notes_loaded() -> None:- global _loaded_notes_run_dir # noqa: PLW0603- run_dir = _get_run_dir()- run_dir_key = str(run_dir.resolve()) if run_dir else "__no_run_dir__"- if _loaded_notes_run_dir == run_dir_key:- return- _notes_storage.clear()- notes_path = _get_notes_jsonl_path()- if notes_path:- _notes_storage.update(_load_notes_from_jsonl(notes_path))- try:- for note_id, note in _notes_storage.items():- if note.get("category") == "wiki":- _persist_wiki_note(note_id, note)- except OSError:- pass- _loaded_notes_run_dir = run_dir_key---def _sanitize_wiki_title(title: str) -> str:- cleaned = "".join(ch.lower() if ch.isalnum() else "-" for ch in title.strip())- slug = "-".join(part for part in cleaned.split("-") if part)- return slug or "wiki-note"---def _get_wiki_directory() -> Path | None:- try:- run_dir = _get_run_dir()- if not run_dir:- return None- wiki_dir = run_dir / "wiki"- wiki_dir.mkdir(parents=True, exist_ok=True)- except OSError:- return None- else:- return wiki_dir---def _get_wiki_note_path(note_id: str, note: dict[str, Any]) -> Path | None:- wiki_dir = _get_wiki_directory()- if not wiki_dir:- return None- wiki_filename = note.get("wiki_filename")- if not isinstance(wiki_filename, str) or not wiki_filename.strip():- title = note.get("title", "wiki-note")- wiki_filename = f"{note_id}-{_sanitize_wiki_title(str(title))}.md"- note["wiki_filename"] = wiki_filename- return wiki_dir / wiki_filename---def _persist_wiki_note(note_id: str, note: dict[str, Any]) -> None:- wiki_path = _get_wiki_note_path(note_id, note)- if not wiki_path:- return- tags = note.get("tags", [])- tags_line = ", ".join(str(tag) for tag in tags) if isinstance(tags, list) and tags else "none"- content = (- f"# {note.get('title', 'Wiki Note')}\n\n"- f"**Note ID:** {note_id}\n"- f"**Created:** {note.get('created_at', '')}\n"- f"**Updated:** {note.get('updated_at', '')}\n"- f"**Tags:** {tags_line}\n\n"- "## Content\n\n"- f"{note.get('content', '')}\n"- )- wiki_path.write_text(content, encoding="utf-8")---def _remove_wiki_note(note_id: str, note: dict[str, Any]) -> None:- wiki_path = _get_wiki_note_path(note_id, note)- if not wiki_path:- return- if wiki_path.exists():- wiki_path.unlink()--def _filter_notes(category: str | None = None,tags: list[str] | None = None,search_query: str | None = None,) -> list[dict[str, Any]]:- _ensure_notes_loaded()filtered: list[dict[str, Any]] = []for note_id, note in _notes_storage.items():if category and note.get("category") != category:@@ -220,9 +68,6 @@ def _to_note_listing_entry("created_at": note.get("created_at", ""),"updated_at": note.get("updated_at", ""),}- wiki_filename = note.get("wiki_filename")- if isinstance(wiki_filename, str) and wiki_filename:- entry["wiki_filename"] = wiki_filenamecontent = str(note.get("content", ""))if include_content:entry["content"] = content@@ -234,16 +79,14 @@ def _to_note_listing_entry(return entry-def _create_note_impl( # noqa: PLR0911+def _create_note_impl(title: str,content: str,category: str = "general",tags: list[str] | None = None,) -> dict[str, Any]:- """Create one note. Public — used by tests."""with _notes_lock:try:- _ensure_notes_loaded()if not title or not title.strip():return {"success": False, "error": "Title cannot be empty", "note_id": None}if not content or not content.strip():@@ -276,13 +119,8 @@ def _create_note_impl( # noqa: PLR0911"updated_at": timestamp,}_notes_storage[note_id] = note- _append_note_event("create", note_id, note)- if category == "wiki":- _persist_wiki_note(note_id, note)except (ValueError, TypeError) as e:return {"success": False, "error": f"Failed to create note: {e}", "note_id": None}- except OSError as e:- return {"success": False, "error": f"Failed to persist wiki note: {e}", "note_id": None}else:return {"success": True,@@ -314,7 +152,6 @@ def _list_notes_impl(def _get_note_impl(note_id: str) -> dict[str, Any]:with _notes_lock:try:- _ensure_notes_loaded()if not note_id or not note_id.strip():return {"success": False, "error": "Note ID cannot be empty", "note": None}note = _notes_storage.get(note_id)@@ -340,7 +177,6 @@ def _update_note_impl() -> dict[str, Any]:with _notes_lock:try:- _ensure_notes_loaded()if note_id not in _notes_storage:return {"success": False, "error": f"Note with ID '{note_id}' not found"}note = _notes_storage[note_id]@@ -355,13 +191,8 @@ def _update_note_impl(if tags is not None:note["tags"] = tagsnote["updated_at"] = datetime.now(UTC).isoformat()- _append_note_event("update", note_id, note)- if note.get("category") == "wiki":- _persist_wiki_note(note_id, note)except (ValueError, TypeError) as e:return {"success": False, "error": f"Failed to update note: {e}"}- except OSError as e:- return {"success": False, "error": f"Failed to persist wiki note: {e}"}else:return {"success": True,@@ -372,19 +203,13 @@ def _update_note_impl(def _delete_note_impl(note_id: str) -> dict[str, Any]:with _notes_lock:try:- _ensure_notes_loaded()if note_id not in _notes_storage:return {"success": False, "error": f"Note with ID '{note_id}' not found"}note = _notes_storage[note_id]note_title = note["title"]- if note.get("category") == "wiki":- _remove_wiki_note(note_id, note)del _notes_storage[note_id]- _append_note_event("delete", note_id)except (ValueError, TypeError) as e:return {"success": False, "error": f"Failed to delete note: {e}"}- except OSError as e:- return {"success": False, "error": f"Failed to delete wiki note: {e}"}else:return {"success": True,@@ -405,10 +230,9 @@ async def create_note() -> str:"""Document an observation, finding, methodology step, or research note.- Notes are your **shared run memory** — they're visible to every- agent in the same scan and persist to ``run_dir/notes/notes.jsonl``- (replayable event log). Wiki-category notes are additionally- rendered as Markdown under ``run_dir/wiki/<slug>.md``.+ Notes are visible to every agent in the same scan for the lifetime+ of the run; they live in-memory only and are cleared when the+ process exits.For actionable tasks, use ``todo`` instead — notes are for capturinginformation, todos are for tracking work.@@ -422,9 +246,7 @@ async def create_note(useful for the final scan report.- ``questions`` — open questions / things to come back to.- ``plan`` — multi-step plans you want to track.- - ``wiki`` — repository or target source maps shared across agents- in the same run. Use this for codebase architecture notes the- whole agent tree should see.+ - ``wiki`` — long-form repository or target maps.Tags are free-form (e.g. ``["sqli", "auth", "critical"]``) — usefulfor later ``list_notes(tags=...)`` filtering.@@ -528,7 +350,7 @@ async def update_note(@function_tool(timeout=30)async def delete_note(ctx: RunContextWrapper, note_id: str) -> str:- """Delete a note. For wiki notes, also removes the rendered Markdown file.+ """Delete a note.Args:note_id: Note id to delete.