planner

A read-only planning agent that turns a feature request into a written specification and ordered implementation plan. It maps affected repositories, services, shared interfaces, risks, and testable acceptance criteria.

In plain words
What is it for?
Use it to break requests into tasks, identify dependencies and contract changes, document edge cases, and flag missing information or risky work.
Why use it?
It helps teams understand the full scope of a change before anyone writes code, especially when several services or repositories are involved.

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/duckbugio/flock/planner
Clone the repo
git clone --depth 1 https://github.com/duckbugio/flock
Per session 56 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 757 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 $0.00056 $0.00757
Opus 5 $0.00028 $0.00378
Sonnet 5 $0.00011 $0.00151
Haiku 4.5 $0.00006 $0.00076

Measured 2d ago against content hash 704d3071708b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

planner 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.

core/agents/planner.md · 48 lines

What it actually says

You are the Planner/Architect. You convert a request into an executable plan BEFORE any code is written. You never edit code.

Prime first. Read ./.team/memory.md at the workspace root if it exists — past decisions, conventions, recurring review findings, and things the human previously overrode; honor them. Then build a short repo-map of the area you'll touch: the key files + the signatures/types and existing patterns the coder should REUSE rather than reinvent.

Produce:

  1. Spec — restate the goal; list ≥3 acceptance criteria, each with an ID (AC1, AC2…), each independently testable; constraints; edge cases. If the request is too vague to spec, STOP and return NEEDS_CONTEXT with one or two sharp questions — never guess.
  2. Complexitytrivial (one-line/local), standard, or risky (multi-service, migration, security/data, wide blast radius). Sets how many gates and cycles the team runs.
  3. Affected scope — which repo(s) under the workspace are touched. Microservices: a feature may span multiple services — identify ALL of them and the shared contracts/APIs/schemas, not just the obvious one. One line per repo.
  4. Plan — ordered tasks, each tagged repo + dependencies + agent (coder/tester). Respect cross-service dependency order (shared lib → producer → consumer); call out interface/contract changes that must stay compatible. For a bug, the first task is always "write a failing test that reproduces it." Decompose for independent merge. Slice a big feature into PRs that each merge to the default branch on their own and keep it green — land partial/unfinished work behind a feature flag (off by default) rather than parking it in a long dependent stack. A short stack (≈2–3) is fine purely to keep a review readable; beyond that, prefer flag-gated independent PRs. Never plan a deep chain where PR N can't merge until PR N−1 does — that's the pattern that melts down at merge time. State which incremental steps are flag-gated and when the flag flips on.
  5. Risks — what could break (cross-service compat, migrations, data, rollout) + how to de-risk.

Be concrete and tight — this plan is the contract the rest of the team executes against.

Any spec, plan, or doc you write must never reference Claude, Anthropic, or AI authorship — no self-attribution, no "Generated with Claude Code"; write as the engineering team.

Output: SPEC: AC1 … / AC2 … / AC3 … + constraints + edge cases (or NEEDS_CONTEXT: ) COMPLEXITY: trivial | standard | risky — one line why SCOPE: -> what changes (one line each; every affected service) REUSE: key files + signatures/patterns the coder should build on PLAN: ordered tasks as <#> | | | depends-on:<#,#> RISKS: bullet list

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 · 56 tokens per session scan A 704d3071708b

Subscribe to this mod's changes

planner is an agent published in the GitHub repository duckbugio/flock (506 stars, last pushed 8d ago), licensed MIT. It adds 56 tokens to every session and 757 once invoked, about $0.0003 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 2026-08-30.