Skip to content

majiayu000/spellbook / skills/threads

threads

Verified

Use when the user explicitly asks for $threads, Codex-native subagents, 开几个子 agent, or a GitHub issue/PR queue needing parallel lanes, worktrees, review/merge gates, and closure audit. Do not use for OS/language threads, chat/email/forum threads, or Assistants product threads unless Codex workflow orchestration is explicit.

FileKindSize
SKILL.mdskill md33.8 KB
agents/openai.yamlconfig671 B
evals/evals.jsonconfig5.7 KB
references/prompt-patterns.mdmarkdown18.7 KB
references/run-log.mdmarkdown16.3 KB
scripts/analyze_run_log.pyexecutablescript8.9 KB
scripts/append_run_log.pyexecutablescript4.0 KB
scripts/run_budget_contract.pyscript17.2 KB
scripts/run_log_schema.pyscript27.5 KB

evals/evals.json

View on GitHub
[
  {
    "id": 1,
    "name": "explicit-codex-native-subagents",
    "should_trigger": true,
    "expected_mode": "execute_direct",
    "expected_output": "Uses the threads skill, records capability and dispatch gates, creates a lane map with disjoint writable files, and does not route to tmux or OS-level threads.",
    "category": "positive-native-dispatch",
    "files": [],
    "query": "Use threads to split this repository cleanup into independent Codex subagents with separate file ownership and review gates."
  },
  {
    "id": 2,
    "name": "github-queue-sweep",
    "should_trigger": true,
    "expected_mode": "execute_direct",
    "expected_output": "Starts with live queue gate requirements, maps issues to PRs, requires GraphQL reviewThreads, and bounds the queue tranche before implementation.",
    "category": "positive-github-queue",
    "files": [],
    "query": "Use threads to process this repo's open GitHub issues and PRs, including review-thread blockers and merge readiness."
  },
  {
    "id": 3,
    "name": "review-only-pr",
    "should_trigger": true,
    "expected_mode": "review_only",
    "expected_output": "Plans a read-only reviewer or merge-reviewer lane, checks current head, CI, merge state, and thread-aware review-thread state, and does not edit files.",
    "category": "positive-review-only",
    "files": [],
    "query": "Open a read-only threads review for PR #42 and tell me whether it is safe to merge."
  },
  {
    "id": 4,
    "name": "research-spec-lanes",
    "should_trigger": true,
    "expected_mode": "research_spec",
    "expected_output": "Splits read-only research lanes by angle, records evidence expectations, and synthesizes without assigning overlapping writable workers.",
    "category": "positive-research-spec",
    "files": [],
    "query": "Use threads to research this agent workflow from product, engineering, and security angles, then synthesize a spec."
  },
  {
    "id": 5,
    "name": "chinese-subagent-request",
    "should_trigger": true,
    "expected_mode": "review_only",
    "expected_output": "Recognizes the explicit subagent request, uses native Codex threads when available, and assigns read-only reviewer lanes with merge-gate evidence.",
    "category": "positive-native-dispatch",
    "files": [],
    "query": "开几个子 agent 分别看这个 PR 的测试、架构和安全问题,最后汇总能不能合并。"
  },
  {
    "id": 6,
    "name": "os-threading-question",
    "should_trigger": false,
    "expected_mode": "single_agent",
    "expected_output": "Answers as an ordinary programming/concurrency question without loading the threads workflow orchestration skill.",
    "category": "negative-language-concurrency",
    "files": [],
    "query": "Explain how Python threads differ from multiprocessing for CPU-bound code."
  },
  {
    "id": 7,
    "name": "rust-thread-bug",
    "should_trigger": false,
    "expected_mode": "single_agent",
    "expected_output": "Treats this as a Rust concurrency debugging task and does not create Codex lane maps or GitHub queue gates unless separately requested.",
    "category": "negative-language-concurrency",
    "files": [],
    "query": "My Rust worker threads deadlock when two mutexes are acquired in opposite order. Help me debug the code."
  },
  {
    "id": 8,
    "name": "forum-thread-summary",
    "should_trigger": false,
    "expected_mode": "single_agent",
    "expected_output": "Treats thread as a discussion artifact, not Codex subagent orchestration.",
    "category": "negative-discussion-thread",
    "files": [],
    "query": "Summarize this long forum thread and extract the main arguments."
  },
  {
    "id": 9,
    "name": "email-thread-reply",
    "should_trigger": false,
    "expected_mode": "single_agent",
    "expected_output": "Handles email writing directly without capability gates, queue gates, or native subagent dispatch.",
    "category": "negative-discussion-thread",
    "files": [],
    "query": "Draft a concise reply to this email thread and keep the tone professional."
  },
  {
    "id": 10,
    "name": "assistants-api-thread",
    "should_trigger": false,
    "expected_mode": "single_agent",
    "expected_output": "Routes to OpenAI API documentation/help rather than Codex workflow orchestration.",
    "category": "negative-product-api-thread",
    "files": [],
    "query": "How do I create and retrieve threads in the OpenAI Assistants API?"
  },
  {
    "id": 11,
    "name": "ambiguous-one-file-task",
    "should_trigger": false,
    "expected_mode": "single_agent",
    "expected_output": "Uses a simple single-agent edit path because the user explicitly rejects agent dispatch and the task is tiny.",
    "category": "negative-simple-task",
    "files": [],
    "query": "Fix this one typo in README.md; no agents needed."
  },
  {
    "id": 12,
    "name": "github-queue-plan-only",
    "should_trigger": true,
    "expected_mode": "plan_only",
    "expected_output": "Runs queue gate and lane planning only, classifies PRs and issues, and stops before edits, comments, pushes, or merges.",
    "category": "positive-plan-only",
    "files": [],
    "query": "Use threads to map all open PRs and issues, but do not edit or merge anything yet."
  },
  {
    "id": 13,
    "name": "large-homogeneous-record-batch",
    "should_trigger": true,
    "expected_mode": "research_spec",
    "expected_output": "Validates a preflight budget, calibrates one tranche with measured usage, projects total calls and tokens, requests confirmation for the concrete budget, and uses a fresh child per tranche without adjacent runtime remediation.",
    "category": "positive-cost-boundary",
    "files": [],
    "query": "Use threads to classify 8,000 similar session records. Keep going until all records are done, and fix any runtime blocker you encounter."
  }
]
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall