v1.7.1 — Zero-owner-data Canvas preview

FindMate v1.7.1 adds one fully fabricated Founder Complement Canvas to the portable Agent Skill. Agents and owners can inspect the exact private renderer output before supplying owner evidence, installing anything, joining a pool, or authorizing a public action.

The checked-in preview is verified byte-for-byte against the current renderer. It contains no real owner data, contact route, proof URL, consent field, or raw note. Unknown evidence remains unknown rather than being labeled a weakness.

Supply chain:

• exact source: 58eb502
• deterministic archive: 18 explicitly allowlisted files
• archive size: 171,976 bytes
• SHA-256: 06e445ceba097f43ac85a2c4cdde741d8ba74a1e23cd83775f03c6f7c729f686
• 63 repository tests and the canonical skill validator passed

Release availability is not evidence of an install, owner opt-in, profile submission, match, endorsement, or star.

v1.7.0 — Private Founder Complement Canvas

Private value before network effects

FindMate v1.7.0 adds the offline Founder Complement Canvas. Agents can turn the
private assessment JSON into a human-readable Markdown result before any public
profile, pool submission, optional repository support, or candidate contact:

python3 scripts/private_report.py owner-assessment.private.json \
  --output founder-complement-canvas.private.md

The canvas:

• summarizes demonstrated 0→1, 1→10, 10→100, and functional contribution
  with evidence counts and confidence;
• separates stated work preferences from demonstrated contribution;
• shows the complement sought and treats missing evidence as unknown;
• includes explicit owner review steps and scientific/identity limits;
• omits raw private notes, proof URLs, contact routes, and consent data;
• performs no network request or public action;
• requires a .private.md filename and is written with mode 0600.

The private assessment now retains its already privacy-checked one-line summary
so the canvas can explain the working hypothesis without copying raw evidence.

Supply chain

• Source commit: 060be6c8dad6f55bcbb377f3b61a83eeadfb1b73
• Portable archive: 17 explicitly allowlisted public files
• Archive size: 169,433 bytes
• Archive SHA-256: 09c77f52d187f97583216b4be2fbd73617ba1e8e79745214d391921de669a7b9
• 62 tests passed; the canonical skill validator and deterministic archive
  verification passed.

Installation, assessment, publication, repository starring, contact, identity
exchange, and introduction remain separate owner-controlled actions.

FindMate v1.6.1 — evidence-calibrated complementarity brief

This documentation release makes FindMate’s research basis easier to audit and harder to overstate.

• adds a focused, primary-source complementarity evidence brief
• corrects the systematic-review DOI and author attribution
• labels the 0→1 / 1→n stage shorthand as a non-psychometric planning aid
• records explicit research boundaries, non-claims, and a prospective outcome-evaluation plan
• changes no matching algorithm, profile schema, GitHub Action, consent gate, or publication behavior

Validation: 56 Python tests; production site build; 3 rendered/discovery tests; lint; production dependency audit with 0 vulnerabilities.

Portable skill archive: 16 allowlisted files, 154,483 bytes, SHA-256 acee70cd816b64c0a5302424ca44d50adff097b48764ff8ec37310e65233369d.

FindMate v1.6.0 — composable validation outputs

The reusable zero-network GitHub Action now exposes two bounded outputs after the full profile check passes:

• canonical_sha256: deterministic hash of the validated public profile
• expires_on: validated YYYY-MM-DD expiry

No alias, summary, evidence, contact route, or other profile field is added to GitHub Action outputs. The existing optional local card draft remains owner-controlled and is never committed or published by FindMate.

The portable Agent Skill archive is unchanged byte-for-byte from v1.5.0 because this release changes only the repository-level Action integration. The archive is deterministic, allowlisted, and accompanied by its SHA-256 file.

Verification: 54 Python tests, two GitHub-hosted Action/validator jobs, two site jobs, the Node 22.14 production build, rendered discovery tests, lint, dependency audit, and deterministic archive verification all passed.

FindMate v1.5.0 — owner-controlled profile cards

Owner-controlled profile-card drafts

The reusable offline Action can now render a deterministic, privacy-minimized Markdown card after the canonical profile validator passes:

- uses: merc1305/findMate@v1.5.0
  with:
    profile: owner-profile.public.json
    card-output: findmate-owner.card.md

The optional step writes only the caller-selected workspace path. It does not commit, upload, publish, message, star, or contact anyone. The owner must inspect and separately approve the exact card and destination before another step publishes it.

The card omits contact routes and raw evidence while retaining an expiry, canonical profile hash, and FindMate attribution. The growth loop stores only the aggregate public marker count and discards code-search items and owner-level data.

Running the Action is not consent to assess an owner, publish a profile or card, star the repository, contact a candidate, or exchange identities.

The deterministic portable skill archive is byte-identical to v1.3.6 and v1.4.0 because the canonical skill contents did not change.

FindMate v1.4.0 — reusable offline validation

Reusable offline profile validation

Downstream repositories can now validate an owner-approved FINDMATE_OWNER_PROFILE_V1 file with the canonical zero-network validator:

- uses: merc1305/findMate@v1.4.0
  with:
    profile: owner-profile.public.json

The composite action passes the caller-selected path as quoted data after --, requests no write permission, and makes no network request. It fails on malformed, unsafe, unconsented, or expired profiles.

The growth loop records only the aggregate count of public workflow references; it does not store repository names, owner identities, workflow runs, or profile data. FindMate is directly reusable as an action and is not represented as a GitHub Marketplace listing.

Running the action or installing the attached skill is not consent to assess an owner, publish a profile, star the repository, contact a candidate, or exchange identities.

The deterministic portable skill archive is byte-identical to v1.3.6 because the canonical skill contents did not change.

FindMate v1.3.6

Portable distribution release: one canonical project-scope Agent Skill plus a deterministic 16-file ChatGPT upload archive and SHA-256 checksum. Installation authorizes no owner assessment or public action. The release asset is 153,911 bytes with SHA-256 6eb080e2ac06b6a15caccbfa2c882394149b7ce9823f649bfaed0bc9b5ab09f6. Maintainer build and verification are not counted as adoption.

What's Changed

• Record G35 release and deployment by @merc1305 in #50
• Record AAS Core G35 readiness by @merc1305 in #51
• Add portable skill distribution by @merc1305 in #52

Full Changelog: v1.3.5...v1.3.6

v1.3.5

What's Changed

• Record protected v1.3.4 skill release by @merc1305 in #41
• Prepare Claude community marketplace review by @merc1305 in #42
• Add no-install private draft entry by @merc1305 in #43
• Record no-install private value deployment by @merc1305 in #44
• Add agent-readable web discovery by @merc1305 in #45
• Record agent-readable discovery deployment by @merc1305 in #46
• Record external discovery verification by @merc1305 in #47
• Record unattributed star increase by @merc1305 in #48
• Add single-comment GitHub profile publication by @merc1305 in #49

Full Changelog: v1.3.4...v1.3.5

v1.3.4

What's Changed

• Record search metadata release by @merc1305 in #36
• Monitor GitHub skill search indexing by @merc1305 in #37
• Record public FindMate launch surfaces by @merc1305 in #38
• Track AAS Core catalog distribution by @merc1305 in #39
• Fix private draft and profile validation by @merc1305 in #40

Full Changelog: v1.3.3...v1.3.4

v1.3.3

What's Changed

• Record licensed native skill release by @merc1305 in #29
• Add native Claude Code marketplace distribution by @merc1305 in #30
• Record Claude marketplace launch by @merc1305 in #31
• Monitor passive distribution surfaces by @merc1305 in #32
• Record monitor launch and draft Moltbook V5 by @merc1305 in #33
• Draft bounded Moltbook Agent Skills launch by @merc1305 in #34
• Improve cofounder search discovery by @merc1305 in #35

Full Changelog: v1.3.2...v1.3.3