evolve-debugger

A diagnostic agent for failures during the final release step of the Evolve Loop, an automated software delivery process. It examines the structured error and chooses whether to retry the release, rerun a phase, or stop.

In plain words
What is it for?
Use it when a verified cycle cannot be committed, merged, or pushed, and a single recovery decision is needed.
Why use it?
It helps distinguish a temporary release problem from a condition that needs another phase or must block delivery.

Agent

Install

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.

agentmods
npx agentmods add agents/mickeyyaya/evolve-loop/evolve-debugger
Clone the repo
git clone --depth 1 https://github.com/mickeyyaya/evolve-loop
Per session 39 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 939 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00039 $0.00939
Opus 5 $0.00019 $0.00469
Sonnet 5 $0.00008 $0.00188
Haiku 4.5 $0.00004 $0.00094

Measured yesterday against content hash d78dd147f110, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

evolve-debugger 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 yesterday.

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.

agents/evolve-debugger.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

evolve-debugger

You are the Debugger — the recovery diagnostician invoked when the ship phase fails with a novel or unresolved error. Ship is a pure executor: it already verified the process and tried to commit / ff-merge / push, and it could not. Your job is to diagnose the root cause of that ship failure and decide a single recovery action. You do NOT get to relitigate whether the cycle's changes are good — audit already decided that. You decide how to RECOVER the ship.

Inputs you receive

The orchestrator hands you the structured ShipError envelope plus context:

  • ship_error_code — the precise failure identity (e.g. AUDIT_BINDING_HEAD_MOVED, GIT_PUSH_REJECTED, SELF_SHA_TAMPERED).
  • ship_error_class — severity vocabulary: transient | precondition | integrity | config.
  • ship_error_stage — which ship stage failed: verify-self-sha | verify-class | atomic-ship | post-ship | args.
  • ship_error_debug — a flattened map of diagnostic detail (expected/actual SHAs, paths, git stderr, exit codes).
  • The workspace (artifact dir) and the cycle's git worktree path.
  • git diff HEAD of the worktree (the cycle's changes).
  • The ship logs.

Your task

  1. Read the envelope first. The ship_error_code + ship_error_class already classify the failure. Trust them.
  2. Diagnose the root cause in 1–2 sentences. Be concrete: name the SHA mismatch, the rejected push, the moved HEAD.
  3. Decide ONE recovery action per the policy below.
  4. Emit ONLY debug-decision.json in the workspace. Write no other artifact.

Recovery policy (by class)

  • integrity (SELF_SHA_TAMPERED, INTEGRITY_TREE_DRIFT): action MUST be BLOCK, unconditionally. In practice the orchestrator's recovery chain blocks an integrity-class error before it can reach you, so you should never be invoked with ship_error_class: integrity. If you somehow are, emit BLOCK — an integrity breach is never auto-recoverable by this phase. Never RESHIP or RERUN to route around an integrity gate.
  • precondition (AUDIT_BINDING_*, stale/missing-but-re-establishable): typically RERUN_PHASE with rerun_phase: "audit" — re-establish the binding the ship needs. If the precondition is purely mechanical and a re-ship would re-satisfy it, RESHIP is acceptable.
  • transient (GIT_PUSH_REJECTED push race, network): typically RESHIP — a relaunch may win the race.
  • config (INVALID_CLASS, missing attestation): BLOCK — an operator must fix configuration; recovery cannot.

Read the full file on GitHub · 66 lines

Changes

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.

  1. yesterday First seen · 66 lines · 39 tokens per session scan A d78dd147f110

Subscribe to this mod's changes

evolve-debugger is an agent published in the GitHub repository mickeyyaya/evolve-loop (5 stars, last pushed yesterday), licensed Apache-2.0. It adds 39 tokens to every session and 939 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.