Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/mickeyyaya/evolve-loop/evolve-diagnose-referencegit clone --depth 1 https://github.com/mickeyyaya/evolve-loopWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00043 | $0.02713 |
| Opus 5 | $0.00022 | $0.01357 |
| Sonnet 5 | $0.00009 | $0.00543 |
| Haiku 4.5 | $0.00004 | $0.00271 |
Grade A, and why
evolve-diagnose-reference scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 246 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Evolve Diagnose Reference (Layer 3 — on-demand)
Layer-3 on-demand reference. Adapted from mattpocock's 6-phase diagnose loop (mattpocock.com / Total TypeScript). Read this when a phase agent or orchestrator is facing a recurring failure and needs a systematic approach to trace and isolate the root cause. When diagnosing a WARN or FAIL cycle, come here before writing guesses into the retrospective. Do NOT read this file on every cycle — it is on-demand only.
Contents
- When to use this reference
- Phase summary
- Phase 1: Observe
- Phase 2: Hypothesize
- Phase 3: Instrument
- Phase 4: Reproduce
- Phase 5: Isolate / bisect
- Phase 6: Fix, verify, remove tags
- [DEBUG-XXXX] Tag Convention
- Evolve-Loop Examples
When to use this reference
| Trigger | Action |
|---|---|
| Audit returns FAIL and the root cause is unclear | Start at Phase 1 — do not skip to a fix |
Same WARN class appears in 3+ consecutive recentFailures entries |
Start at Phase 2 — symptom is already captured |
subagent-run.sh exits non-zero and stderr is empty |
Start at Phase 3 — instrument before re-running |
ship.sh exits non-zero with an audit SHA mismatch |
Use Example 2 below as the instrumentation template |
| Cycle cost is $0.00 and no ledger entry was written | Start at Phase 3 with the sandbox EPERM template (Example 3) |
Phase summary
| # | Phase | Core action |
|---|---|---|
| 1 | Observe | Capture the exact symptom: observed behavior vs. expected behavior. Record the precise error message, exit code, artifact path, and stack trace line. Do NOT hypothesize here. |
| 2 | Hypothesize | Generate 3–5 candidate root causes. Rank by likelihood. Write them down before testing any. |
| 3 | Instrument | Add [DEBUG-XXXX] tagged log/assertion points to test the top hypothesis. Instrument to test a specific hypothesis — NOT to "see what happens." |
| 4 | Reproduce | Reliably reproduce the failure with minimum steps. No fix is valid unless the failure can be triggered on demand. |
| 5 | Isolate / bisect | Eliminate half the candidate causes per iteration. Binary search to the smallest reproduction case. |
| 6 | Fix + verify + remove tags | Apply fix. Verify against the reproduction case. Search grep -r '\[DEBUG-XXXX\]' and remove all diagnostic artifacts before committing. |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 246 lines · 43 tokens per session scan A 6c59891f8db5
evolve-diagnose-reference is an agent published in the GitHub repository mickeyyaya/evolve-loop (5 stars, last pushed 2d ago), licensed Apache-2.0. It adds 43 tokens to every session and 2,713 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
eval-curator
Authors and maintains the brooks-lint eval suite in evals/evals.json — the benchmark scenarios covering R1–R6 (code decay) and T1–T6 (test decay), including the false-positive / tradeoff cases that must NOT be flagged. Ensures every new risk code or skill gets paired coverage and that the suite passes npm run evals.…
ring:streaming-reviewer
Conditional Gate 8 specialist for lib-streaming, business events, outbox, event producers, broker publishing, CloudEvents, and event manifests/catalogs.
lead
Workflow orchestrator. Use for 5-phase TDD coordination, approval gate enforcement, cross-agent task assignment, and phase transitions.
replanner
Triggered by failure-classifier on F2-F4 escalations. Proposes plan-tree mutations: re-decompose stories, mark tasks discarded, re-prioritize children, or promote a node up a tier. Read-only on code; mutations applied via master-planner.
Business & Transformation Master
Genre master for business parables, self-help, and transformation guides - leads practical wisdom books while coordinating with the Author Team.
Visionary Manifestos Master
Genre master for inspirational manifestos on systems building, AI-human partnership, and humanity's golden age - leads paradigm-shifting thought leadership.