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 skills add SeongwoongCho/adaptive-harness --skill task-taxonomygit clone --depth 1 https://github.com/SeongwoongCho/adaptive-harnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/seongwoongcho/adaptive-harness/task-taxonomy)<a href="https://agentmods.dev/skills/seongwoongcho/adaptive-harness/task-taxonomy"><img src="https://agentmods.dev/badge/skills/seongwoongcho/adaptive-harness/task-taxonomy/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/seongwoongcho/adaptive-harness/task-taxonomy"><img src="https://agentmods.dev/badge/skills/seongwoongcho/adaptive-harness/task-taxonomy.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00028 | $0.01917 |
| Opus 5 | $0.00014 | $0.00958 |
| Sonnet 5 | $0.00006 | $0.00383 |
| Haiku 4.5 | $0.00003 | $0.00192 |
Grade A, and why
task-taxonomy 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 10d 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.
Task Taxonomy Reference
Purpose
Adaptive-harness classifies every incoming task on 6 independent axes before routing to a harness. This taxonomy drives the routing decision and ensemble trigger logic. The router agent (LLM-based) reasons over the task description and codebase context to assign values — it does not use keyword heuristics.
The 6 Axes
Axis 1: task_type
What category of work does this task represent?
| Value | Description | Example tasks |
|---|---|---|
bugfix |
Fix a defect in existing behavior | "Login fails when email has uppercase letters" |
feature |
Add new functionality | "Add dark mode support" |
refactor |
Restructure code without changing behavior | "Extract the auth logic into a separate module" |
research |
Investigate, experiment, or benchmark | "Compare three approaches to rate limiting" |
migration |
Upgrade dependencies or migrate between systems | "Migrate from Express 4 to Fastify" |
benchmark |
Measure and analyze performance | "Profile the query execution and find bottlenecks" |
incident |
Urgent production issue requiring fast response | "API returning 500s in production right now" |
greenfield |
Build a multi-component system from scratch | "Build a FastAPI backend with webhooks, InfluxDB, and Grafana dashboard" |
review |
Review code, designs, plans, or PRs | "Review this PR before merging", "Does this design look correct?" |
ops |
QA testing, quality validation, or retrospective | "Run acceptance tests on the new feature", "Conduct a sprint retrospective" |
release |
Release and ship a versioned artifact | "Cut the 2.0.0 release", "Run the ship workflow for this tag" |
Axis 2: uncertainty
How well-defined is the solution before implementation begins?
| Value | Description | Signs |
|---|---|---|
low |
Clear solution path; solution is known before starting | Spec exists, existing pattern to follow, fix is obvious |
medium |
Solution approach is known but details require exploration | Roughly know what to do, some unknowns in implementation |
high |
Solution path is unclear; multiple approaches plausible | Novel problem, no precedent, multiple competing designs |
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.
- 10d ago First seen · 164 lines · 28 tokens per session scan A ace837a459f8
task-taxonomy is a skill published in the GitHub repository SeongwoongCho/adaptive-harness (8 stars, last pushed 5mo ago), licensed MIT. It adds 28 tokens to every session and 1,917 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.
Other skills, from other repositories
relocate-claude-vm
Move the Claude Desktop "Computer Use" sandbox VM bundle (typically 11-13 GB at %APPDATA%\Claude\vmbundles on Windows) off the system drive to a roomier disk via a directory junction. Transparent to Claude Desktop, reversible. Use when the system drive is low on space and a disk scan shows vmbundles as a top consumer.…
autoresearch
Autonomous experiment loop: modifies code, runs experiments, evaluates metrics, keeps improvements. Inspired by karpathy/autoresearch + pi-autoresearch + autoexp. Triggers: /autoresearch, 'auto research', 'optimize continuously', 'experiment loop', 'autonomous optimization'.
security
Claude-driven security audit with an independent second opinion. Integrates the ralph-security agent (6 quality pillars, OWASP A01-A10). Uses LSP for code navigation during analysis. Use when: (1) /security is invoked, (2) task relates to security functionality.
task-batch
Autonomous batch task execution with PRD parsing, task decomposition, and continuous execution until all tasks complete. Uses /orchestrator internally. Stops only for major failures (no internet, token limit, system crash). Use when: (1) processing task lists autonomously, (2) PRD-driven development, (3) batch feature…
diagram-design
Create technical and product diagrams — architecture, flowchart, sequence, state machine, ER / data model, timeline, swimlane, quadrant, nested, tree, layer stack, venn, pyramid — as standalone HTML files with inline SVG. Ships with a neutral editorial skin and a first-run gate that prompts users to customize the…
create-task-batch
Interactive wizard to create PRD or task lists for /task-batch. Uses /clarify and /ask-questions-if-underspecified for precise task definition. Use when: (1) preparing batch execution, (2) creating PRDs, (3) defining task lists with dependencies. Triggers: /create-task-batch, 'create tasks', 'new batch', 'prepare PRD'.