Miosa-osa/OSA
64.3
Adequate · 17 November 2026
419.7k
lines of production code
Elixir
with Rust, Python
1
measurement over time
What this system is
Behavioural changes
1 commit (0 fixes) modifying bench/osabench, bench/osabench/fixtures/acme
A change to existing behaviour in bench/osabench, bench/osabench/fixtures/acme — 1 commit, 55 files: “v1.0.204: regulation layer, per-step routing and advisor, repo map, prefetch, spot-check audits, learning loop, osa.bench and /trace (\#284)” (refs \#284).
bench/osabench, bench/osabench/fixtures/acme · medium confidence · unverified
1 commit (0 fixes) modifying canopy, canopy/backend
A change to existing behaviour in canopy, canopy/backend — 1 commit, 133 files: “feat: add canopy backend, agent dispatch research, and new lib modules”.
canopy, canopy/backend · medium confidence · unverified
1 commit (0 fixes) modifying native/linux/ScreenShare, scripts/windows
A change to existing behaviour in native/linux/ScreenShare, scripts/windows — 1 commit, 15 files: “feat(opencomputers): native desktop consent and cross-platform host foundations (\#275)” (refs \#275).
native/linux/ScreenShare, scripts/windows · medium confidence · unverified
10 commits (3 fixes) modifying bench
A change to existing behaviour in bench — 10 commits (3 fixs), 6 files: “perf: OSA cost 12-35x the field; every instrument that should have shown it was broken”; “feat(bench): scale to 40 hard instances, and fix what the audit found”; “feat(bench): one entry point and a findings queue for the benchmark loop”; and 7 more (refs \#284).
bench · medium confidence · unverified
11 commits (5 fixes) modifying priv/rust/tui/src/style
A change to existing behaviour in priv/rust/tui/src/style — 11 commits (5 fixs), 2 files: “fix(tui): Shift+Enter newline + light-blue thinking verb + hide 0-token count”; “fix(tui): restore OSA's blue identity (revert Claude-orange accents + banner)”; “feat(tui): replies readable while streaming, and a roster that reads as a hierarchy; v1.0.53”; and 8 more (refs \#259, \#255).
priv/rust/tui/src/style · medium confidence · unverified
19 commits (4 fixes) modifying python
A change to existing behaviour in python — 19 commits (4 fixs), 13 files: “feat(bench): run\_bench drops the tasks it cannot grade, and says the denominator first”; “docs(bench): the live docs link to a closed door, and the 5-trial rule is real”; “feat(bench): a reporting gate that refuses to mislead”; and 16 more.
python · medium confidence · unverified
2 commits (0 fixes) modifying osa-screen-capture-windows
A change to existing behaviour in osa-screen-capture-windows — 2 commits, 3 files: “feat(native): Windows Desktop Duplication helper skeleton + Elixir integration”; “feat(opencomputers): native desktop consent and cross-platform host foundations (\#275)” (refs \#275).
osa-screen-capture-windows · medium confidence · unverified
2 commits (0 fixes) modifying priv/macos
A change to existing behaviour in priv/macos — 2 commits, 3 files: “feat(native): PORT= announcement + auto-deploy in macOS/Windows helpers”; “feat(updater): OSA self-update with SHA256 verification + staged swap”.
priv/macos · medium confidence · unverified
2 commits (0 fixes) modifying priv/scripts
A change to existing behaviour in priv/scripts — 2 commits, 2 files: “release: v1.0.84 - a 1M-context model was being treated as a toy”; “feat: tool harness v2 — structured layout, lazy loading, parallel dispatch, frontend pairing”.
priv/scripts · medium confidence · unverified
2 commits (0 fixes) modifying scripts/whatsapp-bridge
A change to existing behaviour in scripts/whatsapp-bridge — 2 commits, 1 file: “chore: merge origin/main (74 commits) — sync with upstream”; “feat: WhatsApp, Matrix, Email, and LINE channel adapters”.
scripts/whatsapp-bridge · medium confidence · unverified
2 commits (1 fix) modifying native/osa\_nif
A change to existing behaviour in native/osa\_nif — 2 commits (1 fix), 2 files: “release: v1.0.77 - the TUI keeps its layout, and three ways to lose data”; “Refactor: restructure core modules, fix onboarding persistence, clean up naming”.
_native/osa\nif · medium confidence · unverified
2 commits (1 fix) modifying workspace
A change to existing behaviour in workspace — 2 commits (1 fix), 1 file: “Refactor: restructure core modules, fix onboarding persistence, clean up naming”; “Update README for current system state, remove loose dev files”.
workspace · medium confidence · unverified
21 commits (7 fixes) modifying desktop
A change to existing behaviour in desktop — 21 commits (7 fixs), 150 files: “feat: complete end-to-end command system for desktop”; “chore: merge origin/main (74 commits) — sync with upstream”; “fix: SSE event types for frontend, expanded spinner verbs”; and 18 more.
desktop · medium confidence · unverified
211 commits (98 fixes) modifying priv/rust/tui/src/app
A change to existing behaviour in priv/rust/tui/src/app — 211 commits (98 fixs), 41 files: “feat(tui): bound inline roster (cap 8 + more-line), clamp N/cap gauge, align fleet-select enter gate”; “feat(tui): fix render duplication, recap suppression+spacing, v1.0.003 padding, token-usage billing; + Ctrl+B background-a-running-command”; “fix(tui): status-bar folder now matches the banner (real working dir)”; and 208 more (refs \#266, \#289, \#104, \#193, \#197, …).
priv/rust/tui/src/app · medium confidence · unverified
23 commits (6 fixes) modifying priv/rust/tui/src/config
A change to existing behaviour in priv/rust/tui/src/config — 23 commits (6 fixs), 3 files: “feat(tui): fix render duplication, recap suppression+spacing, v1.0.003 padding, token-usage billing; + Ctrl+B background-a-running-command”; “feat: CC-parity permission engine + real interrupt/steer (WS3+WS5)”; “fix(security,update): harden git against hostile repo config; verified updates; v1.0.46”; and 20 more (refs \#259, \#292, \#132).
priv/rust/tui/src/config · medium confidence · unverified
26 commits (10 fixes) modifying bench/terminalbench
A change to existing behaviour in bench/terminalbench — 26 commits (10 fixs), 27 files: “feat(bench): run\_bench drops the tasks it cannot grade, and says the denominator first”; “chore(bench): the analysis tooling the stopped arm produced”; “feat(bench): D8 ATIF export, with the argument-fidelity gate that keeps it honest”; and 23 more.
bench/terminalbench · medium confidence · unverified
27 commits (8 fixes) modifying priv/rust/tui/src/tools
A change to existing behaviour in priv/rust/tui/src/tools — 27 commits (8 fixs), 15 files: “style(tui): CC-parity transcript rendering (tool headers, shell output, diffs, thinking)”; “feat(tui): the hunk separator says how much it skipped”; “v1.0.202: see what OSA is doing, stop runaway thinking, parallel read-only shell, settings reach live turns (\#282)”; and 24 more (refs \#282, \#236, \#259).
priv/rust/tui/src/tools · medium confidence · unverified
277 commits (69 fixes) modifying (repo-wide)
A change to existing behaviour in (repo-wide) — 277 commits (69 fixs), 19 files: “chore(release): v1.0.139”; “Enforce cloud-only delegation and explicit stop-all cancellation”; “chore(release): 1.0.102”; and 274 more (refs \#293, \#266, \#229, \#288, \#151, …).
(repo-wide) · low confidence · unverified
28 commits (12 fixes) modifying priv/rust/tui/src/components/agents
A change to existing behaviour in priv/rust/tui/src/components/agents — 28 commits (12 fixs), 3 files: “feat(tui): bound inline roster (cap 8 + more-line), clamp N/cap gauge, align fleet-select enter gate”; “feat: subagent resume/notifications + command surface & deferred sweep (WS7+WS16)”; “feat(flow): live task list + plan round-trip + background-agent dashboard”; and 25 more (refs \#180, \#224, \#259, \#255, \#181, …).
priv/rust/tui/src/components/agents · medium confidence · unverified
3 commits (0 fixes) modifying bench/report
A change to existing behaviour in bench/report — 3 commits, 6 files: “feat(bench): a reporting gate that refuses to mislead”; “perf: OSA cost 12-35x the field; every instrument that should have shown it was broken”; “feat(bench): airgap the agent, and report the honest number”.
bench/report · medium confidence · unverified
31 commits (10 fixes) modifying priv/rust/tui/src/render
A change to existing behaviour in priv/rust/tui/src/render — 31 commits (10 fixs), 13 files: “feat(tui): the hunk separator says how much it skipped”; “test(tui): prove code blocks survive quantization”; “release: v1.0.83 - the composer froze while a turn streamed”; and 28 more (refs \#236, \#259, \#249).
priv/rust/tui/src/render · medium confidence · unverified
32 commits (16 fixes) modifying bin
A change to existing behaviour in bin — 32 commits (16 fixs), 4 files: “fix(release): keep Cargo.lock in step with the version bump”; “fix(tui): responsive status and Claude thinking (\#115)”; “feat(providers): run through the MIOSA AI Gateway from sandbox env (\#290)”; and 29 more (refs \#115, \#290, \#101, \#236, \#249, …).
bin · medium confidence · unverified
4 commits (1 fix) modifying rel
A change to existing behaviour in rel — 4 commits (1 fix), 2 files: “perf(daemon): turn BEAM scheduler busy-wait off so an idle OSA costs \~nothing”; “harden(critical-flows+cross-platform): 47 findings, 7 blockers, 0-failure suite + Windows”; “Refactor: restructure core modules, fix onboarding persistence, clean up naming”; and 1 more (refs \#292).
rel · medium confidence · unverified
4 commits (2 fixes) modifying bench/headtohead
A change to existing behaviour in bench/headtohead — 4 commits (2 fixs), 80 files: “fix(bench): our comparison flattered OSA by construction”; “perf: OSA cost 12-35x the field; every instrument that should have shown it was broken”; “feat(bench): head-to-head harness comparison, model held fixed”; and 1 more.
bench/headtohead · medium confidence · unverified
40 commits (20 fixes) modifying priv/rust/tui/src/components/input
A change to existing behaviour in priv/rust/tui/src/components/input — 40 commits (20 fixs), 6 files: “feat: P0 — agent now operates in the user's folder, not the OSA source tree”; “fix(tui): a queued message now says when it will run”; “fix(subagents): task\_wait 10-min timeout on healthy long agents + raw-id visual leaks (\#180)”; and 37 more (refs \#180, \#259, \#181, \#283).
priv/rust/tui/src/components/input · medium confidence · unverified
46 commits (16 fixes) modifying lib/mix
A change to existing behaviour in lib/mix — 46 commits (16 fixs), 17 files: “fix: green CI - warnings-as-errors, format gate, and the PTY layout lane (\#176)”; “chore: merge origin/main (74 commits) — sync with upstream”; “fix(onboarding): /setup parity with first-run wizard, guard Vault, fix doctor model, drop dead /tui route”; and 43 more (refs \#176, \#148, \#196, \#236, \#153, …).
lib/mix · medium confidence · unverified
5 commits (1 fix) modifying priv/repo
A change to existing behaviour in priv/repo — 5 commits (1 fix), 3 files: “chore: merge origin/main (74 commits) — sync with upstream”; “feat(backend): agent loop improvements, session transcripts, and API enhancements”; “feat(memory): persist RAG vectors to SQLite (survive restart, no re-embed)”; and 2 more.
priv/repo · medium confidence · unverified
5 commits (2 fixes) modifying Formula
A change to existing behaviour in Formula — 5 commits (2 fixs), 3 files: “chore: refresh examples to real loader schemas, fix Homebrew formulas, backend-focus dev.sh”; “feat(tui): replies readable while streaming, and a roster that reads as a hierarchy; v1.0.53”; “Add MIOSA computer use support”; and 2 more.
Formula · medium confidence · unverified
5 commits (2 fixes) modifying bench/swebenchpro
A change to existing behaviour in bench/swebenchpro — 5 commits (2 fixs), 19 files: “feat(bench): SWE-bench Pro harness — and the images ship the answer”; “fix(bench): the whole-suite pytest run that could not resolve its own modules”; “feat(bench): SWE-bench Pro results — 9/12, zero harness faults”; and 2 more.
bench/swebenchpro · medium confidence · unverified
553 commits (260 fixes) modifying lib/optimal\_system\_agent
A change to existing behaviour in lib/optimal\_system\_agent — 553 commits (260 fixs), 1183 files: “fix(providers): OpenRouter's reasoning was billed and thrown away”; “fix: let background subagents run for a full working day (\#133)”; “feat(subagents): sliding-window fan-out (C) + orchestrate swarm visibility (B) (\#170)”; and 546 more (refs \#133, \#170, \#293, \#82, \#176, …).
_lib/optimal\_system\agent · medium confidence · unverified
57 commits (19 fixes) modifying priv/rust/tui/src/components/chat
A change to existing behaviour in priv/rust/tui/src/components/chat — 57 commits (19 fixs), 5 files: “style(tui): CC-parity transcript rendering (tool headers, shell output, diffs, thinking)”; “feat: P0 — agent now operates in the user's folder, not the OSA source tree”; “fix(compaction): the compact point is the model's window, not a constant”; and 54 more (refs \#169, \#282, \#236, \#259, \#153, …).
priv/rust/tui/src/components/chat · medium confidence · unverified
57 commits (20 fixes) modifying priv/rust/tui/src/dialogs
A change to existing behaviour in priv/rust/tui/src/dialogs — 57 commits (20 fixs), 38 files: “harden(round2): untouched-surface fixes + copy polish + 2 real bugs from de-flaking”; “feat: MCP management + output limiting, workspace trust dialog (WS14+WS15)”; “fix(tui): inline permission prompt + tight spacing + stop viewport-rebuild stacking”; and 54 more (refs \#104, \#193, \#197, \#250, \#282, …).
priv/rust/tui/src/dialogs · medium confidence · unverified
6 commits (0 fixes) modifying priv/helpers
A change to existing behaviour in priv/helpers — 6 commits, 4 files: “chore: merge origin/main (74 commits) — sync with upstream”; “Release v1.0.181: macOS capture memory guard (\#252) + Surplus Intelligence provider (\#253)”; “chore(priv): add pre-built osa-screen-capture-darwin arm64 binary”; and 3 more (refs \#252, \#253, \#236).
priv/helpers · medium confidence · unverified
6 commits (3 fixes) modifying bench/terminalbench/driver
A change to existing behaviour in bench/terminalbench/driver — 6 commits (3 fixs), 2 files: “fix(bench): our comparison flattered OSA by construction”; “perf: OSA cost 12-35x the field; every instrument that should have shown it was broken”; “fix(bench): every token figure we published was measured against a private definition”; and 3 more.
bench/terminalbench/driver · medium confidence · unverified
7 commits (1 fix) modifying native
A change to existing behaviour in native — 7 commits (1 fix), 69 files: “feat(native): PORT= announcement + auto-deploy in macOS/Windows helpers”; “Release v1.0.181: macOS capture memory guard (\#252) + Surplus Intelligence provider (\#253)”; “feat(native): Windows Desktop Duplication helper skeleton + Elixir integration”; and 4 more (refs \#252, \#253, \#236, \#275, \#249).
native · medium confidence · unverified
78 commits (33 fixes) modifying config
A change to existing behaviour in config — 78 commits (33 fixs), 3 files: “harden(round2): untouched-surface fixes + copy polish + 2 real bugs from de-flaking”; “Enforce cloud-only delegation and explicit stop-all cancellation”; “fix(security): patch 15 vulnerabilities across auth, shell, channels, and tools”; and 75 more (refs \#193, \#197, \#290, \#252, \#253, …).
config · medium confidence · unverified
8 commits (3 fixes) modifying bench/swebench
A change to existing behaviour in bench/swebench — 8 commits (3 fixs), 26 files: “fix(bench): our comparison flattered OSA by construction”; “perf: OSA cost 12-35x the field; every instrument that should have shown it was broken”; “feat(bench): scale to 40 hard instances, and fix what the audit found”; and 5 more.
bench/swebench · medium confidence · unverified
85 commits (30 fixes) modifying priv/rust/tui/src/client
A change to existing behaviour in priv/rust/tui/src/client — 85 commits (30 fixs), 7 files: “feat(tui): fix render duplication, recap suppression+spacing, v1.0.003 padding, token-usage billing; + Ctrl+B background-a-running-command”; “feat(update): notify in-app when a newer release is available”; “feat: P0 — agent now operates in the user's folder, not the OSA source tree”; and 82 more (refs \#104, \#193, \#197, \#115, \#282, …).
priv/rust/tui/src/client · medium confidence · unverified
89 commits (30 fixes) modifying priv/rust/tui/src/components
A change to existing behaviour in priv/rust/tui/src/components — 89 commits (30 fixs), 16 files: “feat(tui): fix render duplication, recap suppression+spacing, v1.0.003 padding, token-usage billing; + Ctrl+B background-a-running-command”; “fix(tui): status-bar folder now matches the banner (real working dir)”; “feat(update): notify in-app when a newer release is available”; and 86 more (refs \#289, \#115, \#169, \#233, \#282, …).
priv/rust/tui/src/components · medium confidence · unverified
9 commits (1 fix) modifying priv/skills
A change to existing behaviour in priv/skills — 9 commits (1 fix), 186 files: “feat(fleet): self-orchestration — worktree isolation, structured reports, finalizer, playbook”; “feat(security): 0-day whitebox analyzer + CVSS/CWE scoring + RoE safety gate (\#148)”; “Fix TUI resize recovery and add native skills (\#109)”; and 6 more (refs \#148, \#109, \#135, \#147).
priv/skills · medium confidence · unverified
Fixes
1 commit (1 fix) fixing bench/modelsweep
A fix in bench/modelsweep — 1 commit (1 fix), 6 files: “fix(bench): our comparison flattered OSA by construction”.
bench/modelsweep · medium confidence · unverified
10 commits (6 fixes) fixing priv/rust/tui/src/view
A fix in priv/rust/tui/src/view — 10 commits (6 fixs), 4 files: “harden(round2): untouched-surface fixes + copy polish + 2 real bugs from de-flaking”; “feat: MCP management + output limiting, workspace trust dialog (WS14+WS15)”; “Fix v1.0.179: backlog cleanup — base URL, unicode ledger, linger race, latent crashes, test flakiness (\#250)”; and 7 more (refs \#250, \#259).
priv/rust/tui/src/view · medium confidence · unverified
2 commits (2 fixes) fixing priv/rust
A fix in priv/rust — 2 commits (2 fixs), 3 files: “Release v1.0.190: enforce the one-chrome invariant, doc-rot lint, fix a lying test (\#264)”; “fix(security,update): harden git against hostile repo config; verified updates; v1.0.46” (refs \#264).
priv/rust · medium confidence · unverified
29 commits (15 fixes) fixing scripts
A fix in scripts — 29 commits (15 fixs), 18 files: “chore(scripts): fix osa-tui launcher path/port, add rebuild-tui helper”; “chore: remove stale root install.sh + orphaned xref\_graph.dot; fix install refs”; “fix(loop): the announcement backstop matched a vocabulary, not a shape”; and 26 more (refs \#252, \#253, \#196, \#275, \#292, …).
scripts · medium confidence · unverified
54 commits (28 fixes) fixing priv/rust/tui/src
A fix in priv/rust/tui/src — 54 commits (28 fixs), 12 files: “feat(tui): fix render duplication, recap suppression+spacing, v1.0.003 padding, token-usage billing; + Ctrl+B background-a-running-command”; “fix(tui): inline permission prompt + tight spacing + stop viewport-rebuild stacking”; “Release v1.0.192: unbreak the Windows build, bound the PTY hang, gate Windows in CI (\#266)”; and 51 more (refs \#266, \#265, \#259, \#255, \#181, …).
priv/rust/tui/src · medium confidence · unverified
Test coverage
450 commits adding/updating tests in test
Test coverage in test — 450 commits (210 fixs), 1014 files: “fix(providers): OpenRouter's reasoning was billed and thrown away”; “fix: let background subagents run for a full working day (\#133)”; “harden(round2): untouched-surface fixes + copy polish + 2 real bugs from de-flaking”; and 446 more (refs \#133, \#208, \#293, \#176, \#266, …).
test · high confidence · unverified
Dependencies
218 commits updating dependencies (18 manifests)
A dependency / build maintenance change in (dependencies) — 218 commits (56 fixs), 18 files: “fix(deps): pin erlexec to 2.0.6 for OTP 26 compat (\#76)”; “chore(release): v1.0.139”; “fix(runtime): boot as root, split the circuit breaker, scope session settings”; and 214 more (refs \#76, \#293, \#266, \#229, \#151, …).
(dependencies) · high confidence · unverified
Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.
How this codebase got here
Baseline
- First survey — no prior run to compare against. CAI 64.3.
Lenses
- Code Health 73.9
- Architecture 96.3
- Maturity 94.3
- Readiness 61.0
- Security 60.5
- Event-Driven 100.0
- Event Sourcing 100.0
- Performance 60.0
Changes since last survey
- 300 commits — 184 feature/other, 116 fixes
By area
- lib/optimal_system_agent — 137 commits
- priv/rust — 73 commits
- (root) — 26 commits
- (repo) — 14 commits
- test/agent — 7 commits
- test/optimal_system_agent — 7 commits
- test/pty — 6 commits
- config/runtime.exs — 4 commits
- priv/prompts — 4 commits
- test/tools — 4 commits
- priv/skills — 3 commits
- .github/workflows — 2 commits
- desktop/src — 2 commits
- test/channels — 2 commits
- docs/RELEASE_SOP.md — 1 commit
- docs/research — 1 commit
- docs/security — 1 commit
- native/macos — 1 commit
- test/providers — 1 commit
- test/security — 1 commit
Notable commits
- fix: Fix Ollama Cloud route recovery (#110)
- fix: Fix TUI resize recovery and add native skills (#109)
- fix: Fix persistent goal controls and wrapped CLI input (#113)
- fix: Fix subagent skill and Ollama model selection (#111)
- fix: Fix v1.0.173: "Worked for" recap now shows the true total task time (#232)
- fix: Fix v1.0.174: live spinner elapsed matches the recap (wall clock) (#233)
- fix: Fix v1.0.175: repair Windows-1252 mojibake (glm/z.ai em-dashes, quotes, bullets) (#234)
- fix: Fix v1.0.178: 10 confirmed blockers from the v1.0.177 review (#249)
- fix: Fix v1.0.179: backlog cleanup — base URL, unicode ledger, linger race, latent crashes, test flakiness (#250)
- fix: Fix v1.0.180: align every provider/model context window; wire z.ai/GLM into the Registry (#251)
- fix: Fix v1.0.182: retry transient empty/5xx provider responses instead of halting (#254)
- fix: Harden v1.0.176: proactive sweep fixes (Ollama data loss, sync mojibake, image-turn memory) (#235)
- fix: Merge pull request #210 from Miosa-osa/fix/compact-aggressive-fold
- fix: Merge pull request #211 from Miosa-osa/fix/test-isolation-208
- fix: Release v1.0.187: goal-loop lockout fix, /goal auto toggle, MCP test isolation (#261)
- fix: Release v1.0.188: /goal clear ghost fix + measured est-tok/s in model picker (#262)
- fix: Release v1.0.190: enforce the one-chrome invariant, doc-rot lint, fix a lying test (#264)
- fix: Release v1.0.191: fix the tmux chrome-fossil bug (the real one) (#265)
- fix: Revert "fix: green CI - warnings-as-errors, format gate, and the PTY layout lane (#176)"
- fix: Revert "test: stop the runtime-routes cleanup racing late writers (#178)"
- …and 280 more
Architecture
- 1 containers · 2 bounded contexts · 0 dependency edges (baseline)
Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.
Survey your own repository
Miosa-osa/OSA was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.
About this page
- The score is its most recent published measurement, taken on 17 November 2026 at a pinned commit. It is not a live figure and does not change until the project is measured again.
- Measured at commit 047b7b7cdbbd59af7efc25c3c3a5522fa7856a00 — the exact code this score is about.
- Scored under rubric-2026.11.4 — the same rubric and the same method as every other entry in this index.
- Measured by Canine Code Analyzer (cca.canine.dev) using cca measure-2026.11.84+preprod-d645ca9bbedd.