docs: sync CLI documentation with actual command surface

[arepageek]

Description

This PR brings the CLI documentation in line with the actual CLI surface. It does three things:

1. Documents 5 stable commands that were not in docs/commands.md at all — import (with all 5 subcommands), add policy-engine, add policy, add runtime-endpoint, and telemetry (with enable/disable/status).
2. Adds missing flags to commands that were already documented but out of date — create, add agent, add evaluator, add online-eval, dev, and run eval.
3. Fixes incorrect command names in docs/PERMISSIONS.md — run batch-eval and run recommend were used where the actual CLI commands are run batch-evaluation and run recommendation.

After these changes, 100% of stable CLI commands are documented or mentioned in docs/ or README.md (preview commands are intentionally not yet documented).

Files changed

• docs/commands.md — most of the work (new sections + flag updates)
• docs/evals.md — kept in sync with commands.md for the add evaluator, run eval, and add online-eval tables
• docs/PERMISSIONS.md — command name corrections

docs/commands.md — new sections

• ### import — top-level flags + 5 subcommands (runtime, memory, evaluator, online-eval, gateway) with examples and per-subcommand flag tables
• ### add policy-engine — Cedar policy engine creation, attach modes
• ### add policy — Cedar policy authoring (file, inline statement, --generate from natural language)
• ### add runtime-endpoint — named endpoints (version aliases) for runtimes
• ### telemetry — anonymous usage analytics: enable, disable, status

docs/commands.md — flag updates to existing sections

• create: clarified --name (it's the agent name, also used as project dir if --project-name is omitted), added --project-name, --session-storage-mount-path, --with-config-bundle, added AGUI to --protocol
• add agent: added AGUI to --protocol, --request-header-allowlist, --session-storage-mount-path, --with-config-bundle
• add evaluator: refactored to support both llm-as-a-judge (default) and code-based evaluator types via --type, --lambda-arn, --timeout, --kms-key-arn
• add online-eval: added --endpoint <name>
• dev: added -b, --no-browser and --no-traces
• run eval: added --endpoint, -A, --assertion, --expected-trajectory, --expected-response
• remove: added policy-engine, policy, and runtime-endpoint to the examples and the consolidated flag table
• update: documented that update cli is the explicit form of update

docs/evals.md

Same add evaluator / run eval / add online-eval table updates as in commands.md, kept consistent across both.

docs/PERMISSIONS.md

• Replace `run batch-eval` with `run batch-evaluation` (9 occurrences) — matches the real CLI command
• Replace `run recommend` with `run recommendation` (5 occurrences) — matches the real CLI command
• Table column alignment in the logging section is regenerated automatically by Prettier after the replacements

How I verified this

• Built the CLI and parsed --help output for every command and subcommand to extract the real flag set.
• Ran a script that scans every code example (agentcore <cmd> ...) across all docs and asserts every flag exists in the CLI — 0 unknown flags.
• Ran a separate script that lists every CLI command and verifies it has either a dedicated section in docs/commands.md or a clear mention in another doc — 0 stable commands missing.
• Preview commands (ab-test, config-bundle, archive, recommendations, promote, run batch-evaluation, run recommendation, stop batch-evaluation) were intentionally left out of this PR.

Related Issue

Closes #

Documentation PR

N/A — documentation changes are included in this PR.

Type of Change

• Bug fix
• New feature
• Breaking change
• Documentation update
• Other (please describe):

Testing

How have you tested the change?

• I ran npm run test:unit and npm run test:integ (unit tests: 3912/3916 — the 4 failures (DeployStatus, LogPanel, SecretInput) are preexisting on main and unrelated to this PR; verified they fail identically on upstream/main without my changes)
• I ran npm run typecheck
• I ran npm run lint (0 errors; 29 preexisting warnings, none in files touched by this PR)
• If I modified src/assets/, I ran npm run test:update-snapshots and committed the updated snapshots (no asset changes)

Checklist

• I have read the CONTRIBUTING document
• I have added any necessary tests that prove my fix is effective or my feature works (N/A — documentation-only change; verified via the audit scripts described above)
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed (every new section includes runnable examples)
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[notgitika]

LGTM thank you!

[notgitika]

LGTM!