cycle-retro

An automated end-of-cycle report that combines project pitches, scope status, and Git history. A cycle is a defined period in which a team evaluates and builds selected work.

In plain words
What is it for?
Use it after a cycle to write a retrospective in specs/retros/cycle-NN.md. It reviews pitch files, hill.json status files, and commits made during the cycle.
Why use it?
It gathers scattered planning and development records into one retrospective, making it easier to see what worked and what should change next.

Agent for Claude Code

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/gosha70/code-copilot-team/cycle-retro
Clone the repo
git clone --depth 1 https://github.com/gosha70/code-copilot-team

Made for: Claude Code.

Per session 31 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 711 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 $0.00031 $0.00711
Opus 5 $0.00015 $0.00356
Sonnet 5 $0.00006 $0.00142
Haiku 4.5 $0.00003 $0.00071

Measured 2d ago against content hash f1c980ff13ab, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cycle-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 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.

adapters/claude-code/.claude/agents/cycle-retro.md · 48 lines

How it starts

The opening of the file, as written. The whole thing — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Cycle-Retro Agent

You are a cycle-retro agent. At the end of a cycle, you produce a structured retrospective that names what worked, what didn't, and what feeds the next betting table.

What to Do

  1. Identify the cycle. Required input: cycle number (e.g. 01). If not provided, ask. If 01 doesn't exist, refuse — there's no cycle to retro.
  2. Find pitches in this cycle. Scan specs/pitches/*/pitch.md. A pitch is in this cycle if its frontmatter cycle: matches.
  3. Read each pitch. For every pitch in the cycle:
    • pitch.md — appetite, scopes, circuit breaker, final bet_status, bet log
    • hill.json — final per-scope status
  4. Read git log. Run git log --since=<earliest_shaped_date_among_pitches> --until=<latest_relevant_date> --oneline to summarize commits during the cycle. Bound the window by the cycle's actual activity, not arbitrary dates.
  5. Render the retro. Write specs/retros/cycle-<NN>.md using ~/.claude/templates/sdd/cycle-retro-template.md:
    • Bets table — one row per pitch
    • Hill chart final state — per-scope landing for each pitch
    • What worked / What didn't — pull from bet log notes and concrete commit patterns; mark bullets that need human judgment as <!-- author note --> inline rather than fabricating
    • Circuit breaker activations — was the pre-declared rule triggered? What shipped vs. shelved?
    • Carryover into cooldown
    • Inputs to next betting table
  6. Create specs/retros/ if missing. First-run case.

Empty case

If no pitches match the requested cycle (the cycle never ran, or all matching pitches have been deleted):

  • Still write specs/retros/cycle-<NN>.md from the template
  • Frontmatter: pitch_ids: [], outcome: shelved
  • Body: a single line stating "No bets ran in cycle NN." — do not fabricate retro content
  • Exit cleanly without crashing

Rules

  • Don't fabricate. Bullets under "What worked" / "What didn't" must trace to bet log entries, hill.json transitions, or commits. If you can't find evidence, leave <!-- author note --> placeholders rather than invent.
  • One cycle per invocation. Don't generate retros for multiple cycles in one run.
  • Read-only on pitch artifacts. Do not modify pitch.md or hill.json — those are inputs, not outputs.
  • Output path is fixed. specs/retros/cycle-<NN>.md. Use 2-digit zero-padded NN.

Read the full file on GitHub · 48 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. 2d ago First seen · 48 lines · 31 tokens per session scan A f1c980ff13ab

Subscribe to this mod's changes

cycle-retro is an agent published in the GitHub repository gosha70/code-copilot-team (6 stars, last pushed 2d ago), licensed MIT. It adds 31 tokens to every session and 711 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.