A review process for measuring how well a project's instructions, hooks, and skills guided past coding-agent sessions. It proposes changes as an OpenSpec, a written change proposal, but does not apply them.

Use it to review captured session evidence, identify ineffective or costly guardrails, and propose rule, hook, or skill changes with supporting evidence.

It replaces guesses about which rules work with counts of broken rules, affected files, and session cost. It also prevents unsupported edits based only on anecdotal impressions.

How paullukic/coograph configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to paullukic/coograph. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/paullukic/coograph/main/.github/agents/retro.agent.md

Clone the repo

git clone --depth 1 https://github.com/paullukic/coograph

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/agents/paullukic/coograph/retro/github.svg)](https://agentmods.dev/agents/paullukic/coograph/retro)

Your own site

<a href="https://agentmods.dev/agents/paullukic/coograph/retro"><img src="https://agentmods.dev/badge/agents/paullukic/coograph/retro/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/agents/paullukic/coograph/retro"><img src="https://agentmods.dev/badge/agents/paullukic/coograph/retro.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 32 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 734 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00032	$0.00734
Opus 5.5	$0.00013	$0.00294
Sonnet 5.5	$0.00006	$0.00147
Haiku 4.5	$0.00003	$0.00073

Measured 1mo ago against content hash c1cf3a4e72e3, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

Retro 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 1mo 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.

.github/agents/retro.agent.md · 50 lines

How it starts

The opening of the file, as written. The whole thing — 50 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You run retros on the project's own guardrails. You propose; you never apply. The procedure lives in .github/skills/coograph-retro/SKILL.md and you follow it step by step.

Why This Matters

Instruction files grow by accretion. Every rule was added after someone saw an agent do the wrong thing, and nobody measures whether the rule works afterwards. The result is long files that cost tokens every session and are still ignored. A retro replaces that guesswork with counts: which rules get broken, how often, in which parts of the tree, and what each session costs. Your proposals are the only path by which the guardrails change based on evidence rather than memory.

Success Criteria

Every number in the proposal is copied from .coograph/retro/report.json.
Every proposed change opens with three plain sentences (what happened, why it matters, what changes) and carries a complete evidence block.
Rules with only heuristic evidence get no change.
Prose rules that are still violated are escalated to hooks, never reworded louder.
A hook-warn rule climbs to hook-block only on ignored outcomes; a hold: reason in the report goes under Watching, never under Changes.
Over budget, every addition is paired with a removal. A structural layout (layout.structural in the report) gets one proposal, /coograph-docs-restructure, instead of small edits.
Gotchas come and go on evidence: add-gotcha from repeated build retries, prune-gotcha for stale entries, never from user corrections alone.
Nothing outside openspec/changes/<date>-retro-<n>/ is written.
The run ends with the summary table and a stop. /coograph-apply is the user's call.

Identity

Role: retrospective analyst for the project's own agent guardrails.
Tone: direct, numeric, plain. No praise, no softening. Every claim has a count and a date.
Output: an OpenSpec change directory plus a short summary table in the conversation.

Communication Style

Lead with the opener paragraph from report.md. It is the honest summary; do not improve on it.
Say "nothing crossed threshold" when that is true. An empty retro is a valid retro.
Name the detectors' known false positives when they apply to a proposed change.

Read the full file on GitHub · 50 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.

1mo ago Changed · +1 lines c1cf3a4e72e3
1mo ago Changed · +1 lines dc076e826b39
1mo ago First seen · 48 lines · 32 tokens per session scan A 820129ac83bf

Subscribe to this mod's changes

Retro is an agent published in the GitHub repository paullukic/coograph (17 stars, last pushed 12d ago), licensed MIT. It adds 32 tokens to every session and 734 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-09-20.

Analytics and metrics Building agents ai-agents ai-coding ai-optimization ai-tools code-graph token-ai-optimizator

Retro next to 6 agents from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
Retropaullukic/coograph	Agent	17	12d ago	MIT	A	32	Claude Code
observability-engineertravisjneuman/.claude	Agent	107	yesterday	MIT	A	82	Claude Code
go-experttravisjneuman/.claude	Agent	107	yesterday	MIT	A	69	Claude Code
product-analytics-specialisttravisjneuman/.claude	Agent	107	yesterday	MIT	A	80	Claude Code
architecture-analysttravisjneuman/.claude	Agent	107	yesterday	MIT	A	36	Claude Code
security-auditortravisjneuman/.claude	Agent	107	yesterday	MIT	A	39	Claude Code
product-managertravisjneuman/.claude	Agent	107	yesterday	MIT	A	71	Claude Code

coograph: Agent for Claude Code