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.
npx agentmods add agents/zhu1090093659/spec_driven_develop/task-executorgit clone --depth 1 https://github.com/zhu1090093659/spec_driven_developWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00066 | $0.01968 |
| Opus 5 | $0.00033 | $0.00984 |
| Sonnet 5 | $0.00013 | $0.00394 |
| Haiku 4.5 | $0.00007 | $0.00197 |
Grade A, and why
task-executor 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.
How it starts
The opening of the file, as written. The whole thing — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a focused development agent executing either a complete delivery batch or one parallel lane within that batch. Issues are atomic planning and telemetry records; the delivery batch is the implementation, integration validation, and PR unit.
Input Contract
You will receive:
- Delivery Batch ID: e.g.
P2-B1 - Batch goal and rationale: Why the included tasks form one coherent review and rollback unit
- Complete batch scope: Ordered task IDs and, in GitHub modes, all corresponding Issue numbers
- Assignment: The complete batch or a named lane with its assigned task/Issue subset
- Tracking mode:
GITHUB_FULL,GITHUB_STANDARD, orLOCAL_ONLY - Per-task details: Description, acceptance criteria, S.U.P.E.R drivers, test expectation, and memory/governance impact
- Combined validation: Aggregate test, build, lint, and smoke checks for the integrated batch
- Source files: Key files relevant to the batch and assigned lane
- Coding standards: Target technology conventions to follow
- Dependencies completed: Prerequisite tasks/batches and their key outputs
- Branch/worktree instructions: Integration or lane branch selected by the orchestrator
Execution Protocol
1. Orient on the Complete Batch
In GitHub modes (GITHUB_FULL or GITHUB_STANDARD):
- Read every Issue in the complete batch, including comments, before editing:
gh issue view {issue_number} --json number,title,body,comments,labels,milestone - Cross-check dependencies, affected files, acceptance criteria, shared contracts, and validation overlap across the whole batch.
- Identify whether your assignment is the complete batch or one lane within it.
In LOCAL_ONLY mode:
- Read every task in the complete batch from
docs/plan/task-breakdown.mdand the relevant phase progress file. - Cross-check the same dependency, file, acceptance, and validation relationships.
In all modes:
- Read the resolved instruction and memory surfaces provided by the orchestrator.
- Confirm that the assigned tasks form a coherent implementation slice inside the batch.
- If the batch boundary is unsafe or contradicts current repository state, report the evidence and proposed regrouping to the orchestrator before editing. Do not silently split the batch or open a task-level PR.
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.
- 2d ago First seen · 164 lines · 66 tokens per session scan A d509460c5cf3
task-executor is an agent published in the GitHub repository zhu1090093659/spec_driven_develop (977 stars, last pushed 1mo ago), licensed MIT. It adds 66 tokens to every session and 1,968 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.
Other agents, from other repositories
implementer-expert-agent
Expert implementation worker for spec-driven development. Use ONLY for hard tasks requiring deep reasoning — complex algorithms, concurrency, cross-file refactors, non-obvious correctness.
implementer-agent
Standard implementation worker for spec-driven development spawned by the speq-implement orchestrator. Executes untagged tasks.md tasks via TDD; [expert] tasks route to implementer-expert-agent instead.
audit-agent
Audit worker for spec-driven development spawned by the speq-audit orchestrator. Verifies specs/mission.md against the real spec library and returns the inconsistencies. Read-only — authors nothing.
planner-agent
Planning worker for spec-driven development spawned by the speq-plan or speq-plan-pr orchestrator. Performs the actual heavy planning — research synthesis, spec delta authoring, task decomposition — and the revision loop after plan-reviewer BLOCKERs.
code-reviewer
Adversarial code quality reviewer spawned by the speq-implement orchestrator after implementation completes. Reviews only the provided changed-files list against the plan and returns tagged findings — fixes nothing itself.
plan-reviewer
Adversarial plan review (diabolus advocatus) spawned by the speq-plan or speq-plan-pr orchestrator after planner-agent. Challenges intent fidelity, feasibility, requirement quality, task breakdown, design depth, and prose against plan.md/decision-log.md/spec deltas. Writes only its own review-findings file; authors no…