fix(copaw): refresh runtime skill projection

[RerankerGuo]

Summary

• restore the CoPaw standard-to-runtime bridge helpers used by the worker workspace
• project Manager-pushed skills into .copaw/workspaces/default/skills and keep projected skill.json entries enabled
• refresh the runtime skill projection whenever pulled skill files change, so hiclaw apply worker --zip updates are visible without restarting the worker
• preserve current main Matrix identity resolution, template defaults, and user-owned agent configuration while applying controller-owned overlays

Fixes #712.

Verification

• PYTHONPATH=copaw/src python3 -m pytest -q copaw/tests/test_bridge.py (32 passed)
• non-bridge CoPaw suite compared against current upstream/main: both produce 119 passed and the same 27 unrelated Matrix/worker-health failures in this local dependency environment
• git diff --check

Rebase note

Rebased onto current main (6f0c7da). The bridge conflict was resolved by retaining the newer AGENTTEAMS_* / HICLAW_* Matrix user-ID fallback while preserving this PR's template-create and controller-overlay model.

CI note

The refreshed matrix has 18 passing checks and 2 skips. Its only failure is
the OpenClaw/OpenClaw SHARD_A test-03 acknowledgement assertion: the reply
was empty, but the task directory was created and the Worker completed the
task; subsequent tests, including test-06, passed. This does not exercise
the CoPaw bridge changed here.

Temporary dependency

• This branch is temporarily stacked on test: make multi-worker bob setup deterministic #985 (78e9385), whose full CI passed all 10 integration matrices.
• Relative to test: make multi-worker bob setup deterministic #985, this PR retains only its original scoped changes; the shared test fix disappears from the diff after test: make multi-worker bob setup deterministic #985 merges.
• Fresh checks on d7a9fdeb passed every build and all 10 integration matrices.

[github-actions]

📊 CI Metrics Report

Summary

Metric	Current	Baseline	Change
LLM Calls	93	81	+12 ↑ +14.8%
Input Tokens	2741708	2803871	-62163 ↓ -2.2%
Output Tokens	15810	16791	-981 ↓ -5.8%
Total Tokens	2757518	2820662	-63144 ↓ -2.2%

By Role

Role	Metric	Current	Baseline	Change
🧠 Manager	LLM Calls	61	68	-7 ↓ -10.3%
	Input Tokens	1960233	2502214	-541981 ↓ -21.7%
	Output Tokens	10595	13725	-3130 ↓ -22.8%
	Total Tokens	1970828	2515939	-545111 ↓ -21.7%
🔧 Workers	LLM Calls	32	13	+19 ↑ +146.2%
	Input Tokens	781475	301657	+479818 ↑ +159.1%
	Output Tokens	5215	3066	+2149 ↑ +70.1%
	Total Tokens	786690	304723	+481967 ↑ +158.2%

Per-Test Breakdown

Test	Mgr Calls	Wkr Calls	Δ Calls	Mgr In	Wkr In	Mgr Out	Wkr Out	Δ Tokens	Trend
02-create-worker	6	0	-6 ↓ -50.0%	165653	0	929	0	-192040 ↓ -53.5%	✅ improved
03-assign-task	10	8	+3 ↑ +20.0%	276525	197622	1668	1456	+3615 ↑ +0.8%	⚠️ regressed
04-human-intervene	13	13	+13 ↑ +100.0%	336718	300782	1877	1486	+207865 ↑ +48.0%	⚠️ regressed
05-heartbeat	11	6	+10 ↑ +142.9%	322538	162911	2136	984	+213317 ↑ +77.5%	⚠️ regressed
06-multi-worker	21	5	-8 ↓ -23.5%	858799	120160	3985	1289	-295901 ↓ -23.1%	✅ improved

Trends

✅ 2 test(s) improved (fewer LLM calls)
⚠️ 3 test(s) regressed (more LLM calls)

---

Generated by HiClaw CI on 2026-07-13 15:23:37 UTC

---

📦 Download debug logs & test artifacts

[RerankerGuo]

@shiyiyue1102 Hi! This PR addresses #712. Just rebased onto current main (was conflicting) — CI is now green (18/18). Restores CoPaw bridge helpers for workspace → runtime workspace materialization; projects skills/ into .copaw/workspaces/default/skills via symlink and refreshes skill.json on every Manager-pushed skill update. Could you take a look when convenient?

[github-actions]

❌ Integration Tests Failed (llm-interaction / mgr=openclaw / wk=openclaw)

Commit: 8cf9bd3
Workflow run: #1597

Test Results

No test output captured.

Debug Log (tail)

No debug logs available.

---

📦 Download full debug logs & test artifacts

[shiyiyue1102]

Refreshing CoPaw runtime skill projection has real value, but this PR currently mixes several unrelated behavioral changes: a large bridge refactor, Manager heartbeat timing, disabling tool and file guards by default, the skill projection itself, and a stacked multi-worker test commit. It also conflicts with main. Please split out the skill-projection fix into a clean branch with focused tests. Security guard defaults and heartbeat cadence need separate design justification and separate review; they should not ride along with this bug fix.