Skip to content

armgit / ruflo

Public Imported from GitHub
0

🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated

docs(adr): ADR-150 notes catch up through iter 82 (iter 83)

ruva4d1878df5c4
diff --git a/plugins/ruflo-metaharness/scripts/smoke.sh b/plugins/ruflo-metaharness/scripts/smoke.sh
index 31dbc0208c..14c915bfc7 100755
--- a/plugins/ruflo-metaharness/scripts/smoke.sh
+++ b/plugins/ruflo-metaharness/scripts/smoke.sh
@@ -191,6 +191,18 @@ grep -q "execCli(\[\s*'-y'\s*,\s*'metaharness@latest'" "$F" 2>/dev/null || \
grep -q "cwd: opts" "$F" || miss="$miss no-cwd-passthrough"
[[ -z "$miss" ]] && ok || bad "$miss"
+step "17z46. ADR-150 notes reflect iters 60-82 (iter 83)"
+miss=""
+ADR="$ROOT/../../v3/docs/adr/ADR-150-metaharness-integration-surfaces.md"
+grep -q "Phase 3 Β§3.1 βœ… iters 33–82" "$ADR" 2>/dev/null || miss="$miss no-iter-82-status"
+grep -q "eighty-two iterations of /loop" "$ADR" 2>/dev/null || miss="$miss no-82-iter-marker"
+grep -q "Iters 60–82 β€” performance / observability / contract hardening" "$ADR" 2>/dev/null || miss="$miss no-60-82-section"
+grep -q "Three-tripwire upstream-contract defense" "$ADR" 2>/dev/null || miss="$miss no-tripwire-section"
+grep -q "Drift-detection autonomous arc (iters 53-79)" "$ADR" 2>/dev/null || miss="$miss no-drift-arc"
+grep -q "Artifact-tracking family (iters 7 + 69 + 82)" "$ADR" 2>/dev/null || miss="$miss no-artifact-family"
+grep -q "Fleet status (post-iter-82)" "$ADR" 2>/dev/null || miss="$miss no-post-82-fleet"
+[[ -z "$miss" ]] && ok || bad "$miss"
+
step "17z45. bench-similarity artifact + summary in CI (iter 82)"
miss=""
W="$ROOT/../../.github/workflows/metaharness-ci.yml"
diff --git a/v3/docs/adr/ADR-150-metaharness-integration-surfaces.md b/v3/docs/adr/ADR-150-metaharness-integration-surfaces.md
index 0bc390c45f..672b549744 100644
--- a/v3/docs/adr/ADR-150-metaharness-integration-surfaces.md
+++ b/v3/docs/adr/ADR-150-metaharness-integration-surfaces.md
@@ -1,7 +1,7 @@
# ADR-150 β€” MetaHarness Integration Surfaces in `npx ruflo`
-**Status**: Implemented (Phase 1 βœ… iters 1–3 Β· Phase 2 βœ… iters 4–32 Β· Phase 3 Β§3.1 βœ… iters 33–59 Β· KRR retrain pending production data Β· Phase 3 Β§3.2-Β§3.5 scoped in [ADR-151](ADR-151-harness-intelligence-layer.md))
-**Date**: 2026-06-16 (revised 2026-06-16 β€” sixty iterations of /loop)
+**Status**: Implemented (Phase 1 βœ… iters 1–3 Β· Phase 2 βœ… iters 4–32 Β· Phase 3 Β§3.1 βœ… iters 33–82 Β· KRR retrain pending production data Β· Phase 3 Β§3.2-Β§3.5 scoped in [ADR-151](ADR-151-harness-intelligence-layer.md))
+**Date**: 2026-06-16 (revised 2026-06-16 β€” eighty-two iterations of /loop)
**Related**: ADR-148 (cost-optimal router lifecycle via `@metaharness/router`), ADR-149 (per-model cost-optimal routing), ADR-026 (3-tier model routing), ADR-097 (federation budget circuit breaker), ADR-124 (optional native dependencies), ADR-144 (agent-authorization-propagation)
**External reference**: [`ruvnet/agent-harness-generator`](https://github.com/ruvnet/agent-harness-generator) β€” the upstream that publishes `metaharness` + `@metaharness/*`. Same author (rUv), explicitly designed around ruflo primitives.
**Research dossier**: published as a gist (linked from the tracking issue) with full graded-evidence sourcing.
@@ -360,6 +360,114 @@ Both still OPEN as of iter 59. Downstream workarounds remain in place.
- Total ruflo-metaharness smoke surface: ~260 assertions
- 33/33 plugin fleet still green
+### Iters 60–82 β€” performance / observability / contract hardening
+
+A second wave of work after iter 59 focused on three orthogonal
+hardening axes: parallelization, anti-regression infrastructure, and
+upstream-contract tripwires.
+
+#### Parallelization sweep (iters 56-59 + 65 + 67)
+- **iter 56** oia-audit's 5 subprocess calls race via Promise.all
+ β†’ 4.59x measured speedup (close to theoretical 5x).
+- **iter 58** drift-from-history's audit-list + oia-audit race
+ β†’ 1.02-1.03x (audit-list's ONNX warmup dominates; race still
+ proves no serial regression).
+- **iter 59** oia-audit emits `timing.{wallMs, sumComponentMs,
+ parallelSpeedup}` so the speedup is observable and gated against
+ silent serial regression.
+- **iter 65** drift-from-history surfaces the same timing fields.
+- **iter 67** `--baseline-file` fast-path: skips audit-list AND
+ audit-trend memory roundtrip β†’ ~19x speedup (1.4s wall on Apple
+ Silicon vs ~26s slow path). Iter 66's `--baseline-key` was the
+ intermediate ~14x step.
+
+#### Three-tripwire upstream-contract defense (iters 12 + 80 + 81)
+| Tripwire | Surface | Iter |
+|---|---|---|
+| `check-metaharness-compat.mjs` | `@metaharness/router` public API | 12 |
+| `check-mcp-scan-format.mjs` | `harness mcp-scan` text-output format | 80 |
+| `check-fingerprint-schema.mjs` | `metaharness score`/`genome` JSON fields | 81 |
+
+Each tripwire runs in CI `metaharness-real-data` job BEFORE the
+roundtrip β€” upstream drift fails with a SPECIFIC error pointing at
+which surface broke, instead of cascading to downstream symptoms.
+
+#### Anti-regression infrastructure
+- **iter 64** rankSeverity + rollup unit tests (22 assertions) β€”
+ iter-63's shared severity util now anchored at primitive level.
+- **iter 72** parseMcpScanText edge cases (19 assertions) β€”
+ iter-50's parser now anchored against silent regex drift.
+- **iter 73** negative-grep guard: anti-mint-as-MCP enforcement.
+- **iter 74** generalized negative guards: from-repo never wrapped;
+ no new static `@metaharness/*` imports outside neural-router.ts.
+
+#### Bug-discovery + fix arc (iters 47-77)
+The roundtrip test surfaced 4 latent bugs that hand-built-fixture
+testing had missed:
+- **iter 47** β€” score/genome schema mismatch (28 iters silent)
+- **iter 49** β€” audit-trend introduced/cleared was dead code
+- **iter 50** β€” fixed via shared parseMcpScanText util
+- **iter 51** β€” proved drift detection fires on mutation, not just
+ self-match
+- **iter 62** β€” extended SEVERITY_RANK; iter-50 parser unlocked
+ warn/critical findings that the rollup ignored
+- **iter 63** β€” consolidated SEVERITY_RANK + safe rankSeverity()
+ across all 3 consumers
+- **iters 76-77** β€” mutation-tested introduced/cleared at all four
+ corners: clear, introduce, swap, dedup.
+
+#### Drift-detection autonomous arc (iters 53-79)
+- **iter 53** β€” drift-from-history one-command primitive
+ composing audit-list + oia-audit + audit-trend
+- **iter 58** β€” parallel batch
+- **iter 65** β€” observable timing
+- **iter 66** β€” `--baseline-key` fast-path (~14x)
+- **iter 67** β€” `--baseline-file` fastest-path (~19x)
+- **iter 68** β€” roundtrip Stage 8 exercises wrapper end-to-end
+- **iter 69** β€” weekly cron AUTO-runs drift detection every Sunday;
+ downloads prior artifact via `gh run download`
+- **iter 70** β€” drift steps use `if: always() && has_prior` so
+ failure-path artifacts (the most valuable) still get uploaded
+- **iter 75** β€” Stage 9 proves the fastpath catches drift, not just
+ self-match
+- **iter 78** β€” `--alert-on-new-severity` orthogonal alert gate
+ (catches "new CRITICAL finding with similarity intact")
+- **iter 79** β€” weekly cron passes `--alert-on-new-severity high`,
+ closing the production wiring
+
+#### Doctor expansion (iters 45 + 52 + 61)
+- iter 45 separates ruflo-side integration health from upstream-dep
+ presence (different remediation paths)
+- iter 52 verifies parseMcpScanText export
+- iter 61 verifies iter-56 async exports (runHarnessAsync /
+ runMetaharnessAsync) β€” missing = oia-audit parallelization breaks
+
+#### MCP-CLI parity (iters 53 β†’ 71)
+All 9 MCP tools have CLI-level flag parity. Iter 71 specifically
+fixed drift-from-history (iter-66/67 flags were CLI-only until then).
+Iter 73's anti-mint negative guard enforces the deliberate
+asymmetry: 10 CLI subcommands, 9 MCP tools (mint cli-only per
+Β§Sandboxing).
+
+#### Artifact-tracking family (iters 7 + 69 + 82)
+| Artifact | Workflow | Iter |
+|---|---|---|
+| `oia-audit-${run_id}` | weekly cron | 7 |
+| `drift-trend-${run_id}` | weekly cron | 69 |
+| `bench-similarity-${run_id}` | per-PR CI | 82 |
+
+All 90-day retention for cross-artifact comparability.
+
+### Fleet status (post-iter-82)
+- Smoke step count: 19 β†’ 85 (+66 invariants since iter 8)
+- MCP tool runtime contract: 117 assertions across 9 tools
+- Real-data roundtrip: 66 pipeline assertions across 12 stages
+- Unit tests: 94 similarity + severity + parser assertions
+- 3 compat tripwires (router API / mcp-scan text format / fingerprint schema)
+- 33/33 plugin fleet still green
+- 4 CI workflows (metaharness-ci, no-metaharness-smoke, oia-audit-weekly,
+ metaharness-real-data sub-job)
+
### Quote architecture invariant β€” no static metaharness imports
The single non-test ruflo source file that statically imports a