Skip to content

feat(srx-ips, srx-mnha-builder): recognize junos-mcp-server with commit confirmed - #94

Merged
fastrevmd-lab merged 1 commit into
mechubsec:mainfrom
jgrizzuti:feat/skills-commit-confirmed
Oct 2, 2026
Merged

fastrevmd-lab merged 1 commit into
mechubsec:mainfrom
jgrizzuti:feat/skills-commit-confirmed

Conversation

@jgrizzuti

Copy link
Copy Markdown
Contributor

Summary

Adds junos-mcp-server with commit confirmed as a third server in the capability mappings of srx-ips and srx-mnha-builder, alongside Juniper junos-mcp-server v1.1.1 and rust-junosmcp.

The server is the jgrizzuti/junos-mcp-server fork (main at 094c320), which is Juniper's code plus the change proposed upstream in Juniper/junos-mcp-server#34. It exposes ten tools:

  • load_and_commit_config always runs a commit check first, and takes dry_run and confirm_timeout_mins.
  • render_and_apply_j2_template takes confirm_timeout_mins.
  • A new confirm_commit tool confirms the pending commit.

Changes

  • Server identification: the server is identified from the tool list (confirm_commit plus confirm_timeout_mins), not its name. Juniper releases that merge fix: close doc-versus-contract drift found by the codex gate #34 will match the same column.
  • Confirm step: both skills warn that on this server, re-sending load_and_commit_config does not confirm. With no diff it commits nothing, and the rollback still fires. The generic "confirm with a second commit" wording would otherwise mislead the agent.
  • srx-mnha-builder:
    • New dry-run and push mappings for this server (load_and_commit_config with dry_run; the J2 tool with confirm_timeout_mins: 10, then confirm_commit).
    • Both Stage 1 nodes are pushed inside one window, because the ICL checks need both nodes.
    • If verification fails while a confirmed commit is pending, the stage is left to roll back rather than pushing an undo.
    • The existing "confirm Stage 2 before the reboot" rule is unchanged.
  • Versions: srx-ips 0.1.1 → 0.1.2; srx-mnha-builder 0.1.0 → 0.2.0.
  • CHANGELOG entry and skills/CHECKSUMS.sha256 updated for the four changed files.

Evidence

Live test on vSRX 24.4R2.21 through the fork, 2026-10-02:

  1. load_and_commit_config with dry_run: true: commit check passed, the diff was returned, and show system commit showed no new commit.
  2. confirm_timeout_mins: 2, not confirmed: show system commit showed commit confirmed, rollback in 2mins and rollback pending. The device rolled back at +2:45 (by root via other), and the diff against the pre-test rollback was empty. This matches the 30–45 s lag already recorded in srx-ips.
  3. confirm_timeout_mins: 2, then confirm_commit: the change was still present after the window.

Validation

  • Passing locally, run on a clean exported tree:
    • check-skill-packages, check-markdown-links, test-markdown-links
    • check-runtime-intake, check-runtime-intake-safety
    • check-catalog-counts, sync-installer-inventory --check, check-shared-schema
    • git diff --check
    • Checksum contents match a fresh gen-checksums.py run.
  • Not run locally: check-inventory, test-inventory and check-installer invoke ./install.sh, which can't run from Windows Python. They fail the same way on unmodified main in that environment, so CI should cover them.
  • just review (the Codex gate) was not run. It needs FWSKILLS_ALLOW_CODEX_REVIEW=1 and sends the diff off-box.

Per CONTRIBUTING, a maintainer merges this PR; the author won't.

🤖 Generated with Claude Code

…it confirmed

Add junos-mcp-server with commit confirmed (jgrizzuti/junos-mcp-server fork,
proposed upstream as Juniper/junos-mcp-server#34) as a third server in both
capability mappings. It is identified by its confirm_commit tool; both skills
warn that re-sending load_and_commit_config does not confirm there.
srx-mnha-builder pushes stages with a 10-minute confirmed commit and lets an
unverified stage roll back.

Field-confirmed on vSRX 24.4R2.21, 2026-10-02.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@fastrevmd-lab
fastrevmd-lab merged commit 0c0ffa2 into mechubsec:main Oct 2, 2026
2 checks passed
jgrizzuti pushed a commit to jgrizzuti/fwskillsshare that referenced this pull request Oct 5, 2026
…als (mechubsec#95)

* docs(csrx-proxmox-deploy): move gotchas to references, add contents

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs(sd-onprem-proxmox-deploy): outline procedure, move gotchas, add verification checklist

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs(srx-ips): move shared commit/verification guidance to references

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs(srx-dynamic-ip-feed): move TLS and auth patterns to references

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs(srx-policy,srx-mnha): move migration workflow and SRG detail to references

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs(parsers): move app-mapping tables to references, fix examples, add contents

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs: self-contained emitter refs, qualified MCP tools, tighter CIS overview

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs(compliance): add contents lists to skills and long references

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs(srx): add contents lists to skills and long references

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs: add contents lists to remaining skills and long references

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* test(evals): add per-skill fresh-session eval scenarios

Two to three scenarios per skill (prompt, input, must, must not), derived
from each skill's own guidance; dev-only, never installed or published.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* test(evals): enforce an eval file for every skill

scripts/check-evals.py requires evals/<skill>.md with >=2 well-formed
scenarios and no orphans; wired into just lint/test with unit tests.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* chore(skills): patch-bump every skill for the best-practice restructure

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* release: v1.10.0

CHANGELOG section for the skill best-practice restructure and per-skill
evals, folding in the unreleased publishing notes. README clone examples,
install.sh default ref and VERSION move to v1.10.0; the tag is cut from
this PR's merge commit.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix(sd-onprem-proxmox-deploy): free-IP check and flavor disk rule

The new verification checklist inverted the predeployment check (an arping
reply means the address is taken) and the gotcha summary said to resize all
three disks; only the data disks grow, disk-0 stays 200 GB. Found by Codex
review of 6b7affa.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix(evals): ground every scenario in its skill's own text

Codex review of f379ed6 found assertions contradicting the skills: a
non-canonical schema in conversion/audit/diff/compliance inputs, SD Cloud/OVA
scenarios for the SD On-Prem skill, a missing STIG fixture, and wrong control
IDs for PCI, HIPAA, CMMC and ISO 27001. All scenarios were re-audited against
SKILL.md and references; unsourced and unobservable assertions removed.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* ci: run the eval checker and its tests on pull requests

skills-validate.yml lists scripts explicitly rather than calling just, so
check-evals.py and test-check-evals.py never ran in CI. Found by Codex review
of 3cdc10f.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix(srx-ips,sd-onprem-proxmox-deploy): restore confirm-commit and HA-cert recovery guidance dropped in merge

The main-branch merge into this restructuring branch kept the pre-mechubsec#94
wording in two places: srx-ips's confirm-commit step lost the pointer to
the server's own confirm step, and sd-onprem-proxmox-deploy's MNHA/cluster
certificate recovery procedure was dropped entirely. Both are restored.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants