slice

A command that runs one implementation slice from planning through delivery. An implementation slice is a bounded piece of project work, and the command delegates its coordination to a delivery lead.

In plain words
What is it for?
Use it to select a slice, confirm its size and prerequisites, create an implementation packet and task ledger, and coordinate the required specialists.
Why use it?
It prevents an oversized or loosely coordinated change from being started without acceptance criteria, dependencies, tasks, and a test plan.

Command 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 commands/jeet129/praxis/slice
Clone the repo
git clone --depth 1 https://github.com/jeet129/praxis

Made for: Claude Code.

Per session 27 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 980 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 $0.00027 $0.00980
Opus 5 $0.00014 $0.00490
Sonnet 5 $0.00005 $0.00196
Haiku 4.5 $0.00003 $0.00098

Measured 2d ago against content hash fa52eb935474, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

slice 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.

.claude/commands/slice.md · 87 lines

How it starts

The opening of the file, as written. The whole thing — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Run ONE implementation slice end-to-end. You do NOT execute the slice yourself. You do NOT spawn a specialist. You delegate the whole thing to delivery-lead, which orchestrates.

What you do

Ask me which slice (if not obvious from the roadmap in .project/working/). Show me:

  • Acceptance criteria
  • Estimated size (must be 2-5 days; if larger, decompose first — reject the slice back to /architect or /discover if it's too big)
  • Stack(s) involved
  • Dependencies on prior slices

Once I confirm, delegate to delivery-lead via the Task tool:

Task({
  subagent_type: "praxis:delivery-lead",
  prompt: "Run implementation-slice.yaml for slice <slice-id>.

  Follow the workflow STRICTLY. Do NOT skip any step.

  1. Open the slice. Update .project/working/active-workflow.md.

  2. Spawn lead-developer FIRST. Give it the slice spec, AC, and touched
     modules. Wait for its output — BOTH artifacts, not just the first:
       - the implementation packet at
         .project/working/slice-<slice-id>-packet.md (decomposed tasks per
         specialist, dependencies, relevant NFRs, test-plan skeleton)
       - the task ledger at .project/working/slice-<slice-id>-tasks.yaml
         (per references/loop-contracts.md §2: per-task agent, tier, ac,
         verify command, depends_on). The ledger is what makes the slice
         drive-eligible — without it /drive has nothing to iterate.

  3. ONLY AFTER the packet + ledger exist, dispatch specialists.
     PREFERRED: lead-developer dispatches them per the ledger's dependency
     DAG (two-tier delegation — you route phases, Lead Developer routes
     tasks). FALLBACK: only if the harness cannot nest agent spawns, you
     dispatch specialists yourself, passing each its portion of the packet.
     Run specialists in parallel where the DAG allows.

  4. When every specialist has produced a PR, spawn (in parallel):
       - code-reviewer for every PR
       - ux-designer for visual review when the slice has UI tasks
         (screenshots vs design plan + tokens, per frontend-design)
       - security-reviewer if the slice is security-bearing
         (touches auth, data-handling, public surface, deps, or
         compliance-bearing code)
       - qa-engineer for every PR

  5. If any reviewer returns a blocker: fix loop. Same specialist that
     owns the file gets the finding; re-review by the blocking reviewer.
     Loop until all reviewers pass.

  6. Close the slice. Write acceptance verdict and episodic entry.
     Update .project/working/active-workflow.md with next-slice pointer
     or workflow-complete marker.

  NEVER spawn a specialist directly for slice implementation. Lead
  Developer builds the packet first. This rule holds even if the slice
  looks small or only one specialist appears needed."
})

Read the full file on GitHub · 87 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 · 87 lines · 27 tokens per session scan A fa52eb935474

Subscribe to this mod's changes

slice is a command published in the GitHub repository jeet129/praxis (7 stars, last pushed 5d ago), licensed MIT. It adds 27 tokens to every session and 980 once invoked, about $0.0001 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.