feat(cli): add manus-agent cwe subcommand for CWE weakness lookup

[manus-use]

Summary

Adds a manus-agent cwe <CWE-ID> CLI subcommand that wires up the existing get_cwe_details Strands tool for direct terminal use.

What was added

• _build_cwe_parser() — argparse setup: positional CWE-ID, --output {text,json}
• _run_cwe(argv) — CLI runner with:
  • Input normalisation: accepts CWE-79, cwe-79, or plain 79
  • Text output: formatted CWE ID, MITRE URL, wrapped description
  • JSON output: structured dict with cwe_id, description, url
  • Error handling: invalid input → exit 1, tool errors → exit 1, import failures → exit 1
• _SUBCOMMANDS registration + main() dispatch wiring
• README.md: TOC entry + CLI reference section between blast-radius and silent-patches
• tests/test_cli_cwe.py: 38 tests across 7 suites (parser, text output, JSON output, errors, registry, dispatch, import path)

Usage

manus-agent cwe CWE-79
manus-agent cwe 787
manus-agent cwe CWE-89 --output json | jq .description

Test results

Full suite: 1196 passed (baseline 1158 + 38 new), 0 failures.

Duplicate check

Checked all 50 open PRs (#51, #53, #54, #58, #60, #64, #65, #67, #74–90, #96, #98, #100, #103–123) and 30 merged PRs — no existing open or merged PR adds a cwe CLI subcommand. PR #89 adds tool-level tests for get_cwe_details but does not wire it to the CLI. PR #112 adds a cvss subcommand (different tool, different purpose).

Mock strategy

All tests patch manus_agent.tools.get_cwe_details.get_cwe_details — no real HTTP calls.