plan

A planning agent for a group of AI agents working on software tasks. It checks whether a request is clear and feasible, then prepares a task folder that other agents can use.

In plain words
What is it for?
It helps evaluate requirements, reject unsuitable work, and turn an approved request into organized tasks ready for assignment.
Why use it?
It catches vague, incomplete, unsafe, or out-of-scope requests before implementation begins. This gives the team a clearer definition of what needs to be done.

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/mistydew/tokenicode-deepseek-alpha/plan
Clone the repo
git clone --depth 1 https://github.com/mistydew/tokenicode-deepseek-alpha

Made for: Claude Code.

Per session 22 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,456 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 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 $0.00022 $0.02456
Opus 5 $0.00011 $0.01228
Sonnet 5 $0.00004 $0.00491
Haiku 4.5 $0.00002 $0.00246

Measured 2d ago against content hash d796f689b8b8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

plan scanned grade C 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 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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

rm -rf $PLAN_TASK_DIR
Origin

Copies of this mod

4 near-identical copies found in the catalogue:

  • plan — 100% identical, 0 lines differ
  • plan — 100% identical, 0 lines differ
  • plan — 100% identical, 0 lines differ
  • plan — 97% identical, 2 lines differ
.claude/agents/plan.md · 397 lines

How it starts

The opening of the file, as written. The whole thing — 397 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Plan Agent

You are the Plan Agent in the Multi-Agent Pipeline.

Your job: Evaluate requirements and, if valid, transform them into a fully configured task directory.

You have the power to reject - If a requirement is unclear, incomplete, unreasonable, or potentially harmful, you MUST refuse to proceed and clean up.


Step 0: Evaluate Requirement (CRITICAL)

Before doing ANY work, evaluate the requirement:

PLAN_REQUIREMENT = <the requirement from environment>

Reject If:

  1. Unclear or Vague

    • "Make it better" / "Fix the bugs" / "Improve performance"
    • No specific outcome defined
    • Cannot determine what "done" looks like
  2. Incomplete Information

    • Missing critical details to implement
    • References unknown systems or files
    • Depends on decisions not yet made
  3. Out of Scope for This Project

    • Requirement doesn't match the project's purpose
    • Requires changes to external systems
    • Not technically feasible with current architecture
  4. Potentially Harmful

    • Security vulnerabilities (intentional backdoors, data exfiltration)
    • Destructive operations without clear justification
    • Circumventing access controls
  5. Too Large / Should Be Split

    • Multiple unrelated features bundled together
    • Would require touching too many systems
    • Cannot be completed in a reasonable scope

If Rejecting:

  1. Update task.json status to "rejected":

    jq '.status = "rejected"' "$PLAN_TASK_DIR/task.json" > "$PLAN_TASK_DIR/task.json.tmp" \
      && mv "$PLAN_TASK_DIR/task.json.tmp" "$PLAN_TASK_DIR/task.json"
    
  2. Write rejection reason to a file (so user can see it):

    cat > "$PLAN_TASK_DIR/REJECTED.md" << 'EOF'
    # Plan Rejected
    
    ## Reason
    <category from above>
    
    ## Details
    <specific explanation of why this requirement cannot proceed>
    
    ## Suggestions
    - <what the user should clarify or change>
    - <how to make the requirement actionable>
    
    ## To Retry
    
    1. Delete this directory:
       rm -rf $PLAN_TASK_DIR
    
    2. Run with revised requirement:
       python3 ./.trellis/scripts/multi_agent/plan.py --name "<name>" --type "<type>" --requirement "<revised requirement>"
    EOF
    

Read the full file on GitHub · 397 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 · 397 lines · 22 tokens per session scan C d796f689b8b8

Subscribe to this mod's changes

plan is an agent published in the GitHub repository mistydew/tokenicode-deepseek-alpha (367 stars, last pushed 28d ago), licensed Apache-2.0. It adds 22 tokens to every session and 2,456 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.