Skip to content

CHANGELOG.md

Latest commit

 

History

History
149 lines (122 loc) · 7.54 KB

CHANGELOG.md

File metadata and controls

149 lines (122 loc) · 7.54 KB
Raw

Changelog

All notable changes to this repo are documented here. Format loosely follows Keep a Changelog.

[3.2.0] — prompt-shape retrofit + full illustrative eval coverage

No skill count change (still 41: 23 stage + 10 initiative + 8 intelligence). Retrofitted all 41 skills to the current prompt shape in 9 batches: bant, improve-demo, deep-discovery; meddicc, meddpicc, faint, sandler, challenger; pain-finder, next-step-commitment, sell-the-dream, first-call-rampup; persona-based-discovery, technical-discovery, meeting-to-qualify, sql-to-qualify, qualify-to-poc; poc-success, pilot-to-conversion, negotiation-closing, new-business-registration, end-client-registration; churn-prevention, customer-success, upsell-expansion, onboarding-journey; prospect-pipeline-check, qualify-pipeline-check, evaluation-pipeline-check, deal-risk-digest; poc-pilot-pipeline-check, negotiation-pipeline-check, won-pipeline-check; mutual-action-plan, champion-tracker, deal-highlights, follow-up-email-drafter; win-loss-brief, executive-briefing, competitor-battlecard, adoption-leaderboard. Documented the gtm-context.md convention and closed the remaining skill gaps. Added illustrative self-authored evals (not Tier 3, see evals/SCORECARD.md) for the last skills still missing them, completing illustrative eval coverage across all three shapes (A, B, C) for all 41 skills.

[3.1.1] — README: drop Coverage table

No skill count change (still 41: 23 stage + 10 initiative + 8 intelligence). Removed README's "Coverage" table (duplicated the Available skills table's Stage/Initiative info with no added signal); "Where this stops" moved down to sit right above FAQ. check-docs-sync.sh and its fixture tests updated to match (one fewer check, one fewer test case).

[3.1.0] — illustrative evals + README rework

No skill count change (still 41: 23 stage + 10 initiative + 8 intelligence). Added illustrative self-authored evals (not Tier 3, see evals/SCORECARD.md) for the 8 remaining Shape B (CSV pipeline-sweep) skills: evaluation-pipeline-check, negotiation-pipeline-check, prospect-pipeline-check, qualify-pipeline-check, poc-pilot-pipeline-check, won-pipeline-check, end-client-registration, new-business-registration. Reworked README's "See it work" section to cover all three output shapes (transcript-audit, CSV pipeline-sweep, document-writer) with real isolated runs and a generalized "try any skill" instruction.

[3.0.0] — 41 skills, deal-intelligence dimension, eval framework

41 Agent Skills (up from 32 at 2.1.0), 23 stage + 10 initiative + 8 intelligence.

New intelligence dimension (8, all new): mutual-action-plan, deal-risk-digest, deal-highlights, follow-up-email-drafter, win-loss-brief, executive-briefing, champion-tracker, competitor-battlecard. These write something forward (a plan, a digest, a drafted email) instead of grading a call against a rubric, so neither stage nor initiative fit — see MAINTAINING.md's "Two dimensions" for the frontmatter contract and the Coverage-table exemption they get.

New initiative skill (1): sell-the-dream, anchored on Andy Raskin's publicly published "5 Elements of a Great Sales Narrative" (no canonical "sell the dream" framework exists — see the skill's provenance note).

Also lands the Tier 0 eval framework (evals/framework/*, isolated RUN/GAP/LEARN stages, judge.py never sees SKILL.md) and evals/gt-web/, a web-sourced ground-truth mechanism using real published call critiques in place of hand-labeled transcripts, since Tier 3 gold-label recall stays blocked on having no in-house labeler. Four illustrative self-authored evals (meddicc, challenger, deal-risk-digest, champion-tracker) ship labeled explicitly as not Tier 3.

[2.1.0] — 32 skills, registration coverage

32 Agent Skills (up from 29 at 2.0.0), 23 stage + 9 initiative.

New stage skills (3): poc-success (a completed POC/pilot's exit review — success criteria verdicts, end-user validation, sponsor sign-off, production-readiness blockers), new-business-registration (a new-logo deal-registration record for completeness), end-client-registration (a partner/reseller's end-client deal registration, kept distinct from direct pipeline).

Also: README's "How these fit together" skill map moved off a Mermaid/d2 diagram to a plain ASCII flowchart in a fenced code block — the rendered image scaled down to unreadable at README width; the fixed-width text version doesn't have that failure mode.

[2.0.0] — 29 skills, ROADMAP.md, dev/main release flow

29 Agent Skills (up from 16 at 1.0.0), 20 stage + 9 initiative, plus the process this repo now ships work through: ROADMAP.md tracks what's built and what's next, work lands on dev and ships to main in batched, fully-gated releases (this entry is the first one cut that way).

New initiative skills (4): meddpicc, faint, sandler, challenger.

New stage skills (9): sql-to-qualify, qualify-to-poc, persona-based-discovery, and the deal-stage pipeline-check family — prospect-pipeline-check, qualify-pipeline-check, evaluation-pipeline-check, poc-pilot-pipeline-check, negotiation-pipeline-check, won-pipeline-check. The pipeline-check six are a new input class for this repo: no transcript mode, they read a .csv deal export and flag deals that can't pass their own stage's exit test, citing the column and cell value behind every flag.

Also ships:

  • ROADMAP.md — every phase-1 skill tracked as a checkbox, machine-checked against skills/*/ by scripts/check-docs-sync.sh's roadmap-in-sync rule so this file can't silently drift the way this changelog once did.
  • dev as the integration branch for skill PRs; main only moves in batched, fully-validated releases via the new gtm-release skill.
  • scripts/check-pr-diff-identity.py — scans every PR's actual diff for the operator's own git identity or a home-directory path right after gh pr create, since the PR is already public by the time a pre-push gate could catch it.

[1.0.0] — Initial public release

16 Agent Skills across two axes — deal stage and cross-stage initiative — each auditing a transcript or CRM export against a GTM rubric with a quote or timestamp behind every finding.

Stage (11): deep-discovery, meeting-to-qualify, technical-discovery, improve-demo, pilot-to-conversion, negotiation-closing, first-call-rampup, onboarding-journey, customer-success, churn-prevention, upsell-expansion.

Initiative (5): meddicc, bant, pain-finder, next-step-commitment, adoption-leaderboard.

A third axis, vertical-context (industry reference packs for cybersecurity, healthcare, fintech), was built but held back from this release pending domain-expert review — see the vertical-context branch.

Also ships:

  • validate-skills.sh, scripts/check-docs-sync.sh, scripts/scan-content.py — zero-dependency local checks, all three run in CI on every PR, backed by a fixture test suite (tests/run-checks-tests.sh) that proves each one actually catches what it claims to.
  • A Claude Code plugin (/plugin marketplace add zime-ai/zime-gtm-skills), alongside clone-and-copy and git-submodule install paths.
  • EVALS.md's three-tier eval methodology, and a worked example comparing a skill-aided run against an unaided one.
  • CONTRIBUTING.md / MAINTAINING.md documenting the full skill-authoring, review, and landing process, and the IP boundary on what's fair to draw from Zime's own product.