retro

A command that reviews recent work using Scrumban, a workflow that combines Scrum planning with Kanban flow. It examines completed and blocked tasks, repeated causes of failure, solution times, bypasses, and possible lessons.

In plain words
What is it for?
Use it to review the last two weeks by default, or another period such as a week, month, quarter, or chosen number of days.
Why use it?
It helps reveal recurring problems, slower delivery, and process rules that people are repeatedly bypassing.

Command

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 commands/kouroshez/coding-os/retro
Clone the repo
git clone --depth 1 https://github.com/kouroshez/coding-os
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 656 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 $0.00000 $0.00656
Opus 5 $0.00000 $0.00328
Sonnet 5 $0.00000 $0.00131
Haiku 4.5 $0.00000 $0.00066

Measured yesterday against content hash 8250cdd37b99, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, 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 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.

src/core/commands/retro.md · 39 lines

What it actually says

Run a Scrumban retrospective for the period specified by $ARGUMENTS (default: 14 days).

Steps:

  1. Resolve period: if $ARGUMENTS is a number → days; if $ARGUMENTS is week|month|quarter → 7|30|90 days; if empty → 14 days.
  2. Call MCP tool cos_task_retro(since="Nd") (canonical aggregation of completed + blocked + failure patterns; since is a duration string like "14d").
  3. Call cos_failure_pattern_query(limit=20) to surface recurring failure modes (frequency-ranked root_cause categories from backtrack_events; treat count ≥ 2 as recurring — filter client-side).
  4. Call cos_metric_trend(metric="time_to_solution", window_days=N) to detect process drift.
  5. Read $COS_PANEL_DIR/.clear1-bypass-log (if present): each line is a self-issued CLEAR 1 enforcement-bypass plus its justification. Report the count + justifications — a rising count means the discipline is being routed around, not internalized.
  6. Promotion candidates: call cos_learn_suggest(limit=20) and filter client-side to Trusted, not yet promoted lessons (confidence ≥ 0.7 AND times_validated ≥ 3; suggest already excludes promoted rows). For each candidate propose a destination (an existing rule file section, or a skill) and show the draft from cos_promote(pattern_id, target)apply NOTHING without explicit user approval. On approval, write the returned content to the agreed destination (governance edit — needs the task marker); the tool already stamped promoted_to, which removes the row from digest/suggest.
  7. Render in the template below. The action items are the only place the agent should propose content beyond the data — every line needs an owner placeholder + due-date placeholder (the user fills them in).

Template:

## Retro — YYYY-MM-DD (period: {N} days)

### What shipped ({count})
- [TASK-NNN] {title} → complete, {date}

### What broke / blocked ({count})
- [TASK-NNN] {title} — blocked for {days}; root cause: {category}

### Patterns (from cos_failure_pattern_query)
- {recurring category}: {N occurrences}; suggested mitigation: {action}

### Trend (time_to_solution)
- median {Nm} (vs {Nm} previous period) — {improving | stable | regressing}

### Self-issued CLEAR-1 bypasses ({count})
- {justification} — what discipline was skipped, and was it warranted

### Promotion candidates (Trusted lessons → durable rules; {count})
- [#{pattern_id}] {lesson one-liner} — validated {N}×; proposed destination: {rule/skill path} — approve?

### Action items
- [ ] {action} — @owner — due YYYY-MM-DD

No-blame rule (per incident-response): describe the failing system, never the person.

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 · 39 lines · 0 tokens per session scan A 8250cdd37b99

Subscribe to this mod's changes

retro is a command published in the GitHub repository kouroshez/coding-os (6 stars, last pushed yesterday), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 656 tokens. 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.