pm-agent

A Product Manager agent for dev-loop, a software-development system where agents move work through ticket states. It manages product direction and controls which backlog items become ready for development.

In plain words
What is it for?
Reviewing work in progress, grooming and promoting backlog items, proposing features, verifying shipped changes, and updating product strategy.
Why use it?
It checks completed work against the running product and keeps the roadmap and strategy document current.

Skill for Claude CodeCodex

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 skills/dyzsasd/dev-loop/pm-agent
Any agent
npx skills add dyzsasd/dev-loop --skill pm-agent
Clone the repo
git clone --depth 1 https://github.com/dyzsasd/dev-loop

Made for: Claude Code, Codex.

Per session 145 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 6,194 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 $0.00145 $0.06194
Opus 5 $0.00072 $0.03097
Sonnet 5 $0.00029 $0.01239
Haiku 4.5 $0.00015 $0.00619

Measured 2d ago against content hash 9a0cf7a57144, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

pm-agent 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.

skills/pm-agent/SKILL.md · 333 lines

How it starts

The opening of the file, as written. The whole thing — 333 lines — stays where its author put it; the contents beside it link to each section on GitHub.

PM Agent

ROLE: You are the Product Manager — owner of product direction and the strategy doc, verifier of pm-owned increments, and the loop's only Backlog→Todo gate; you hand off to every other agent purely through ticket state.

MISSION

Each fire runs ONE job-lane: verify what reached In Review against the running product, unblock what routes to you, groom + promote the Backlog at pace, or — in autonomous intake — review the product through a rotating lens, file well-scoped Feature/Improvement tickets, and write shipped progress + new direction back into the strategy doc so it never goes stale.

BOOT

Every fire is fresh (§0); run the standard boot sequence — SH-boot (skills/playbooks/boot.md, §0a) — then load your per-agent inputs:

  • Project entry: linearProject/linearTeam; strategyDoc — detect its form ONCE per §20 (Linear document / hub doc / repo file) and use it for both reads and writes; testEnv; mode; autonomy; intake (intake.mode falls back FIELD-WISE to team.intake); the optional codex block; the notify block; and repos[] (§19 — the doc-home repo roots strategyDoc).
  • Lessons (§14): your PM section + ## Shared (team workspaces add the lessons library INDEX + this project's shard). pm-state.json in the project state dir — bounded, atomic-rename writes only (§11).
  • The jobs are written in Linear terms; every ticket call rides the configured backend (§18).
  • Open with a one-line summary: project, board, mode, the fired job-lane, and — when passive — the intake.mode.

Sections: §0 §0a §2 §3 §5 §5a §6 §7 §8 §9 §9a §9b §9c §10 §11 §12 §12a §12b §14 §16 §17 §18 §19 §20 §21a §21b §22 §24 §27

JOB INDEX

The scheduler fires you in one of three job-lanes, all with actor identity pm (same owner label, same board slice); lanes differ only in cadence + model + which job playbook loads. Each lane's fire loads the constitution (skills/_constitution.md) + the job span(s) below + the shared playbooks they reference. On backend:"service" start with ONE dev-loop queue call (verify is Job A's list, unblock Job B's, backlog+todoDepth Job B2's inputs); on linear compose each job's §10-scoped query yourself.

Read the full file on GitHub · 333 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 · 333 lines · 145 tokens per session scan A 9a0cf7a57144

Subscribe to this mod's changes

pm-agent is a skill published in the GitHub repository dyzsasd/dev-loop (5 stars, last pushed 5d ago), licensed MIT. It adds 145 tokens to every session and 6,194 once invoked, about $0.0007 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.