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 skills/davidteren/intent-engineering/ie-initnpx skills add davidteren/intent-engineering --skill ie-initgit clone --depth 1 https://github.com/davidteren/intent-engineeringWrote 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/davidteren/intent-engineering/ie-init)<a href="https://agentmods.dev/skills/davidteren/intent-engineering/ie-init"><img src="https://agentmods.dev/badge/skills/davidteren/intent-engineering/ie-init.svg" alt="Measured on agentmods" 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 | $0.00090 | $0.05915 |
| Opus 5 | $0.00045 | $0.02958 |
| Sonnet 5 | $0.00018 | $0.01183 |
| Haiku 4.5 | $0.00009 | $0.00592 |
Grade A, and why
ie-init scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 475 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Intent Engineering — Init / setup / upgrade
Scaffolds and upgrades project config under .intense/ so a team can declare ways of
working (lens toggles, conventions, auto sources, severity align), design-pattern
policy (preferred / allow / block / approved), and architecture thresholds — then
commit them. Once present, .intense/ supersedes the plugin defaults
(${CLAUDE_PLUGIN_ROOT}/references/config-resolution.md) for every ie-* run in that
tree (via walk-up).
Modes (auto-detected when $ARGUMENTS is blank; overridable by token):
| Mode | When | What it does |
|---|---|---|
| fresh | no project .intense/ yet (or user said fresh / all / file tokens) |
Full setup wizard → write files |
| upgrade | .intense/ exists (or user said upgrade) |
Diff capabilities vs defaults; merge only missing keys with confirm |
| calibrate | user said calibrate [percentile] |
Measure distributions; propose thresholds with evidence |
Also supports legacy file tokens: all, ways, patterns, thresholds (skip profile
questions when non-interactive; still run stack-aware copy rules).
Interactive tool
If presenting menus interactively in Claude Code, pre-load AskUserQuestion
(deferred tool): call ToolSearch with select:AskUserQuestion once before the first
menu. If the harness has no blocking-question tool (ToolSearch returns nothing / call
fails), fall back to a numbered list and wait for the user's reply — never silently pick.
Procedure
0. Resolve mode + detect layout (always)
0a. Mode
- If
$ARGUMENTSstarts with tokencalibrate→ calibrate (optionalp90/p75/p95; default p90). Jump to Step 2c. - Else if
$ARGUMENTShas a whole-tokenupgrade(space-separated; not a substring of another word) → upgrade. Ifways/patterns/thresholdsalso appear, limit upgrade merges to those files only. Jump to Step U after a light detect pass (0b–0c still run so the report names stacks/roots). - Else if
$ARGUMENTShasfreshor a file token (all/ways/patterns/thresholds) ormulti-repo/roots:…→ fresh with that selection (multi-repo/roots imply fresh + profile multi-repo; still run 0b–0d). - Else (blank): discover nearest usable
.intense/via walk-up (${CLAUDE_PLUGIN_ROOT}/references/config-resolution.md).- Found → recommend upgrade; still offer fresh re-scaffold (overwrite only with
confirm). Present:
1) upgrade 2) fresh (re-scaffold) 3) calibrate. - Not found → fresh.
- Found → recommend upgrade; still offer fresh re-scaffold (overwrite only with
confirm). Present:
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago First seen · 475 lines · 90 tokens per session scan A 30ed8539909d
ie-init is a skill published in the GitHub repository davidteren/intent-engineering (3 stars, last pushed 18d ago), licensed MIT. It adds 90 tokens to every session and 5,915 once invoked, about $0.0005 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
refactoring-the-assembled-ui
Edit, split or share the CSS and JavaScript under plugins/audit/scripts/ui/ — files Python concatenates into one inline and one inline carrying code, in a self-contained page opened over file://. Covers the assembly contract and the byte-level pins in plugins/audit/tests/ that guard it, why the split must be an…
running-resumable-sync-jobs
Design and review long-running batch or sync jobs that process many items against a remote API and persist checkpoint state — exit codes that report partial failure, checkpoints that are safe to resume, per-item tolerance vs. fatal abort, telling "zero" apart from "couldn't fetch", bounded retries, and honest…
writing-javascript
How browser JavaScript is written in this repo — both surfaces now speak one modern dialect and this says what that means for the code you type, where ES modules are impossible and where they are merely unused, and how to pick a feature (Baseline plus a file:// gate). Covers functional idiom over imperative DOM work…
choosing-what-to-optimize
Decide what is worth making faster in this repo, and prove it paid — measure the thing instead of grepping for it, separate wall clock from total CPU from the critical path, rank targets by shape rather than by how many modules import them, and decline the work with a number when the number says so. Use this whenever…
reproducing-ci-locally
Run the CI gate on your machine so it agrees with the runner — deriving the exact command, paths, markers, and env from the workflow file instead of the Makefile, unblocking gate steps that short-circuit and hide the next failure, pinning the linter version CI resolves, and confirming the run is green instead of…
verifying-external-behavior
Confirms what a third-party library, remote API, build backend, or scraped document actually does before writing code that depends on it — throwaway probes that run in seconds, permissive clients that forward wrong arguments instead of rejecting them, per-endpoint docs that don't generalize, response shapes that make…