v0.6.7

Removed warning from Claude Desktop that was misleading.

Full Changelog: v0.6.6...v0.6.7

v0.6.6

Complete Claude Desktop/Code Support

• Converted all commands into skills
• Simplified skill names where possible
• Reduced front matter tokens to reduce skill tax

Full Changelog: v0.6.5...v0.6.6

Full Changelog: v0.6.5...v0.6.6

v0.6.5

Full Changelog: v0.6.4...v0.6.5

Update to logo handling for the REPS report.

Full Changelog: v0.6.4...v0.6.5

v0.6.4

Rename the scenario-row behavioral-arc field script → guidance

"Script" invited the caller model to recite the field; it actually holds an arc to be pursued.
That mismatch degraded transcript realism — the evidence base every report finding rests on.

Clean cut, no alias. reps/rows.py requires guidance and rejects a legacy script key via a
dedicated ROW_LEGACY_RENAMED_INPUT_FIELDS, checked before the required-field loop so the error
names the rename rather than a confusing "missing guidance". Explicit rejection is required, not
cosmetic: input is an open-ended denylist, so a stale key would otherwise ride along silently.

Scope

• 116 JSONL rows across 12 banks — key only, 0 rows differ by anything else
• 10 drivers — token {scenario_input.guidance}; Behavior arc: label and canonical structure unchanged
• 5 pillar rubric judges — renamed, and no longer describe the caller as "scripted"
• 3 slot templates — {behavioral_arc_directives} → {behavioral_arc_guidance}
• Skills, _meta.md docs, reps/README.md, and the living 002 row contract
• Not migrated: tuned overlays under results/ (321 rows). They fail validation by design; the
  message tells operators to re-run reps-profile.

Also in this branch

README content gaps. Both READMEs advertised "eight skills and six slash commands" — there are
11 and 7, with all three reps-* skills missing from the table. Added the full inventory and
install instructions for Claude Code, Cursor, claude.ai, and the Claude API.

Full Changelog: v0.6.3...v0.6.4

v0.6.3

okareo-tools 0.6.3

Fixed: In keyless (MCP) mode with no local reps/ install, reps-run could produce hand-assembled findings records and reports that didn't match the platform contract. MCP mode now fetches the pure-Python report tooling via get_reps_baseline into a temporary directory and runs capture/render through the canonical code — same records, same report, in every mode.

• Records and reports are now guaranteed generator-produced; if tooling can't be obtained, the run stops with recovery guidance (/okareo:reps) instead of improvising — captured simulation data stays safe on the platform.
• reps-explore's standalone mode is explicitly scoped to its own draft outputs.
• No action needed: existing local installs are unaffected; keyless runs pick up the fix automatically.

Full Changelog: v0.6.2...v0.6.3

v0.6.2

REPS drivers conform to the new Okareo Driver template standard

Rewrite all 10 shipped drivers to the platform's four-section core
(Persona / Scenario Details / Objectives / Soft Tactics) and drop
hand-authored rule blocks, which the MCP now appends canonically.

Capture the appended blocks verbatim in reps/shared/driver-canonical-blocks.md
and append them on the SDK upload path so run_suite.py stores the same
effective prompt as an MCP save (and fingerprints match). Add
tests/test_driver_conformance.py to enforce the shape and prove every
driver still renders against its routed rows.

Per-clause accounting and open judgment calls: specs/014-driver-conformance/
conformance-record.md (review pending; verification sims not yet run).

Full Changelog: v0.6.1...v0.6.2

v0.6.1

Addition of implementation suggestions as part of the reporting workbench results.

Full Changelog: v0.6.0...v0.6.1

v0.6.0

Full Changelog: v0.5.1...v0.6.0

v0.5.1

Direct users to gather tool content from okareo-ai/okareo-tools on Github. This will be superseded by an MCP path in a future release.

Full Changelog: v0.5.0...v0.5.1

v0.5.0

okareo-tools v0.5.0

REPS agent-evaluation workbench
New top-level reps/ directory: a standalone, forkable harness that evaluates any registered agent across four pillars — Reasoning · Execution · Performance · Security — and generates an explainable assessment report. Each pillar ships committed baseline scenarios, drivers, and checks (including a full security attack suite: authority-pressing, goal-hijacking, covert manipulation, verification-gate probes), plus a report generator and a scenario/target reuse layer. Start at reps/README.md.

Three new skills (8 → 11)

• reps-explore — run one benign discovery conversation against an unknown agent and draft its profile
• reps-profile — interview the operator and tune the REPS suite to their agent
• reps-run — execute a pillar against a registered target and regenerate the report
• Automated publishing
• This repo is now published automatically from Okareo's private okareo-tools-dev repository on every change — .sync-state.json at the root fingerprints each sync, and direct edits here are flagged as drift and reconciled upstream. See CONTRIBUTING.md for how to propose changes.

Also in this release

• Refinements to the existing skills (better target/driver/simulation creation guidance)
• Release workflow and build/validate script updates; Okareo logo assets added
• Install / upgrade: /plugin marketplace update okareo-tools then /plugin install okareo@okareo-tools

Full Changelog: v0.4.0...v0.5.0