intent

A command set for managing a change proposal from draft to shipped. It stores each proposal in a folder whose location records its current stage and can also check whether it is ready to implement.

In plain words
What is it for?
Use it to view intent status, create proposals from quoted text, run a readiness check, and conduct a planning interview that produces a detailed specification.
Why use it?
It keeps observations, questions, and planned changes separate, so unfinished ideas are not mistaken for approved work.

Command

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/intentdriven/abcd/intent
Clone the repo
git clone --depth 1 https://github.com/intentdriven/abcd
Per session 37 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,314 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 $0.00037 $0.02314
Opus 5 $0.00018 $0.01157
Sonnet 5 $0.00007 $0.00463
Haiku 4.5 $0.00004 $0.00231

Measured yesterday against content hash 3cdb353cf5e0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

intent 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 yesterday.

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.

commands/intent.md · 188 lines

How it starts

The opening of the file, as written. The whole thing — 188 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/abcd:intent — intent lifecycle

The write side of the intent record store under .abcd/development/intents/. Every intent gets a stable itd-N id and directory-as-truth lifecycle state (drafts/, planned/, shipped/, disciplines/, superseded/). Bare invocation performs zero writes.

Status (bare)

"${CLAUDE_PLUGIN_ROOT}/abcd" intent --json

Summarise the JSON for the user: counts per bucket, open/closed spec counts, and the intent↔spec links. Nothing is created or moved by this invocation.

Which ledger? A half-formed observation, question, or nitpick goes to /abcd:capture "…"; a user-facing change you want to ship goes to /abcd:intent "…". For a big, unproven idea there is an optional third route: /abcd:ideate runs the admission gauntlet and records the verdict either way. It is a pointer, never a precondition — filing a draft without it is a normal thing to do.

Decompose before filing (itd-84, hand-run)

One proposal is rarely one record. Before running the create command below, run the itd-84 decomposition as an advisory analysis the human confirms — capture routes the pieces, it never files a monolith:

  1. Route. Split the proposal into parts and route each to its record home — user-facing capability → an intent; trust-boundary rule → an ADR (plus a brief invariant); standing stance → a principle; plumbing → the brief. Render the result as a table: part | type | home.
  2. Link. Surface the existing records the proposal touches, with typed links — supersedes / reverses / duplicates / refines — never "related". A reversal ("this reverses invariant X") is flagged for the human to confirm, never auto-classified.
  3. Verdict. Propose one of three outcomes — FILE-AS-IS / SPLIT / HOLD. The human adopts the routing; only the part they confirm as an intent proceeds to the create command, and the other parts go to their homes in the same session (or are captured so they are not lost).
  4. Grade. Append the confirmed table — and whether the initial routing survived the human's confirmation — to the dated decomposition-calibration note under the development record's research/notes/. That corpus (about 50 graded captures) is what gates the automated rung.

Read the full file on GitHub · 188 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. yesterday First seen · 188 lines · 37 tokens per session scan A 3cdb353cf5e0

Subscribe to this mod's changes

intent is a command published in the GitHub repository intentdriven/abcd (3 stars, last pushed 2d ago), licensed MIT. It adds 37 tokens to every session and 2,314 once invoked, about $0.0002 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.

Related

Other commands, from other repositories