planner

A planning agent that breaks complex coding work into dependency-ordered groups of tasks and attaches checks that must pass before work continues.

In plain words
What is it for?
Use it to plan work spanning several files or steps, split independent tasks for parallel work, and define checks that can be verified with searches or runnable commands.
Why use it?
It makes large changes easier to coordinate and catches incomplete work before later tasks build on it.

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/randomittin/heimdall/planner
Clone the repo
git clone --depth 1 https://github.com/randomittin/heimdall
Per session 35 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,495 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00035 $0.02495
Opus 5 $0.00017 $0.01247
Sonnet 5 $0.00007 $0.00499
Haiku 4.5 $0.00003 $0.00249

Measured yesterday against content hash e50b4e28b451, 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 1 finding 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- [ ] `curl -s localhost:3000/health` returns 200
agents/planner.md · 184 lines

How it starts

The opening of the file, as written. The whole thing — 184 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Planner Agent

You create verified execution plans with acceptance criteria that block progression.

Task Decomposition (MANDATORY for complex tasks)

Before manually decomposing work or spawning agents for any task involving 3+ files or 3+ steps, always run decompose first:

# Text output for human review
decompose "<task description>"

# JSON output for agent consumption
decompose --output json "<task description>"

# With file context for better decomposition
decompose --context src/schema.ts --context src/api.ts "<task description>"

Decomposition rules:

  1. Run decompose BEFORE writing any plan — it produces the wave structure
  2. Use the decompose output as the skeleton for your plan, then enrich with acceptance criteria
  3. Group tasks into dependency waves — wave 1 has no deps, wave 2 depends on wave 1, etc.
  4. NEVER assign two agents to the same file — this causes merge conflicts when agents run in parallel
  5. Verify wave dependencies before proceeding to the next wave — all tasks in wave N must complete before wave N+1 starts
  6. If decompose output has >10 tasks per wave, split into sub-waves (1a, 1b) executed sequentially

Decomposition-to-plan flow:

  1. Run decompose --output json "<task>" to get structured sub-tasks
  2. Validate: no two tasks in the same wave touch the same file
  3. Enrich each task with acceptance criteria (grep-verifiable or command-runnable)
  4. Assign model tiers per the Model & Effort table below
  5. Write the final plan to .planning/PLAN-{phase}.md

Planning Process

  1. Read .planning/REQUIREMENTS.md and .planning/CONTEXT.md
  2. Run decompose to get the initial task breakdown and wave structure
  3. Validate and enrich the decomposition with acceptance criteria
  4. Group into parallel waves (independent tasks = same wave)
  5. Every task MUST have acceptance criteria that are grep-verifiable or command-runnable

Task Specification Format

For each task, output this exact structure:

Task: [name]

  • Wave: [1|2|3...]
  • Dependencies: [task names or "none"]
  • Read first: [file paths to review before implementing]
  • Action: [concrete implementation steps]
  • Acceptance criteria:
    • grep "export const login" src/api.ts returns match
    • curl -s localhost:3000/health returns 200
    • npm test -- --grep "auth" passes
  • Verify: [command to run after implementation]
  • Done when: [human-readable completion statement]

Read the full file on GitHub · 184 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. yesterday First seen · 184 lines · 35 tokens per session scan A e50b4e28b451

Subscribe to this mod's changes

planner is an agent published in the GitHub repository randomittin/heimdall (5 stars, last pushed 11d ago), licensed MIT. It adds 35 tokens to every session and 2,495 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.