agentsclimarketplace

Lamina init

Skill aryaniyaps/lamina/skills/lamina-init

Headless Product Design skill for AI coding agents | Design how it works, verify what you ship.

Install
npx -y skills add aryaniyaps/lamina --skill lamina-init

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Use only when explicitly invoked as lamina-init. Establish business evidence and canonical Product, Actor, and Persona resources in Lamina's transactional graph.

SKILL.md

4.4 KB, 828 tokens by cl100k_base, as published. Nobody here has run it

/lamina-init

Before any evidence or graph mutation, read and apply ../lamina/orchestrator/prerequisites/cli-required.md.

Read ../lamina/orchestrator/audit-profiles.yaml. Load the init.always references, then only those init.conditional references whose when signal is present. For each selected entry, open the named capability SKILL.md and then its exact reference; do not load sibling topics.

For a directory that is not yet a Git project, initialize Git metadata as specified by that prerequisite before writing evidence. Do not ask for or create an initial commit.

Init writes evidence files only under .lamina/ and canonical knowledge through graphd. It never edits application source.

Evidence artifacts

Write .lamina/business-context.md with frontmatter containing lamina.maturity, platform, and last_updated, followed by exactly these non-placeholder sections: Problem statement, Business goals, Success metrics, Scope, Users & market, Product posture, Constraints, Stakeholders, Risks & unknowns, Research posture, Triad check.

Write .lamina/personas.json as evidence-source JSON with evidence-grounded personas. Goals, constraints, and evidence are arrays. Do not invent demographics. These files are indexable evidence, not canonical graph state.

Run the shipped init/persona validators when available, then lamina graph observe so CocoIndex produces explicit source Observation envelopes. The CLI automatically replaces an incompatible graphd and retries observation once; do not repeatedly rebuild observations to repair daemon compatibility.

If observation exits nonzero after that recovery, treat observation as degraded and optional. Core graph initialization may continue, but do not create Evidence Resources or attach evidence claims for the unavailable snapshot. Do not describe aliases, observation view names, or generation labels as evidence.

When observation succeeds, use observed.resource_ids from its output (or the exact Resource.id values returned by querying that active observation view) as graph evidence. Never substitute a path alias, source key, view name, or generation for an Observation Resource id.

Canonical graph

Start one explicit session. Propose:

  • one inferred Product Resource proposal grounded in the user's explicit product intent;
  • every evidence-grounded Persona as a Persona Resource;
  • corresponding Actor Resources when authority/ownership is known;
  • lamina:canAssume Statements between Personas and Actors;
  • Evidence Resources referencing the relevant source observations.

Publish atomically. Agents must not submit epistemic class or approval. Never cap Personas. Record the actual returned Resource ids for Product, Personas, Actors, and Evidence; generated canonical Resources normally use res_* ids. Do not report input aliases as ids.

All agent-accessible proposal methods use inferred ingress, including claim.add. Never select an epistemic class by choosing a method name. Intended knowledge requires a trusted engine-owned intent ingress; until that ingress supplies it, preserve the user's words as provenance and keep the proposal inferred.

Update mode

Merge changed business evidence, append a dated changelog, rerun observations, and propose new Statements or aliases without replacing stable Resource identity. Preserve conflicting valid facts as Contradictions.

Completion

Report canonical graph initialization and observation as separate outcomes. Always report the GraphVersion, source revision, actual Product/Persona/Actor Resource ids, contradictions, and evidence gaps. Install or refresh passive provider rules with lamina setup --agent <provider>, then state that future ordinary product requests automatically prepare graph-backed implementation context, complete design gaps, implement, and verify. Do not recommend another slash command.

Only report observation coverage and Observation Resource ids when lamina graph observe exited zero and its completion checks passed. After a nonzero observation command, report observation as degraded/unavailable and explicitly state that the published GraphVersion has no current observation-backed evidence. Never claim complete observation coverage on that path.

Gives 0 of the 12 instructions most project setup skills give in 828 tokens

Counted across 999 of the 1,637 authors here whose files we hold, read 2026-08-06

  • ask one question at a timein 29 of 999, across 28 files
  • detect the package manager from lockfilesin 28 of 999, across 9 files
  • present findings to the userin 25 of 999, across 4 files
  • explore current repo statein 24 of 999, across 3 files
  • update the agent skills block in place if it existsin 24 of 999, across 3 files
  • install husky lint-staged and prettierin 23 of 999, across 4 files
  • create the lintstagedrc filein 22 of 999, across 3 files
  • commit all changed filesin 22 of 999, across 3 files
  • run lint-staged to verify it worksin 22 of 999, across 3 files
  • initialize huskyin 21 of 999, across 2 files
  • create the husky pre-commit filein 21 of 999, across 2 files
  • create a prettierrc file if missingin 21 of 999, across 2 files

Said here and by no other author read

  • apply required cli prerequisites before any mutation
  • load only init.always and matching init.conditional references
  • initialize git metadata for non-git project directories
  • write evidence files only under the .lamina directory
  • write business-context.md with specified frontmatter and sections
  • write evidence-grounded personas to personas.json

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.