---

Skjermbilder

---

Detaljer

---

---

ITHZ MCP Hovedfunksjoner

• Project ledger
  Keeps allowed claims, blocked claims, reviewer notes, and replication records explicit and inspectable.
• MCP host installers
  Includes setup paths and sample configs for Codex, Claude Code, Cursor, Antigravity, and generic MCP clients.
• Automatic checkpoints
  Lets agents leave a structured record of what changed, why it changed, what was verified, and what remains next.
• Native archive support
  Works with ITHZ memory zones and native archive workflows for inspectable project memory packaging.
• Large-repository intake
  Prunes generated outputs and noisy folders so large projects can build useful context without unnecessary clutter.
• Inspectable memory trail
  Makes project memory visible and reviewable instead of burying important decisions inside one-off conversations.
• Redaction-aware workflow
  Supports safer memory intake by keeping sensitive or inappropriate content out of durable project records.
• Project-owned agent memory
  Stores decisions, gates, risks, claims, workflow notes, and checkpoints in a local project memory zone.
• Deterministic context packs
  Gives AI agents bounded, relevant project context instead of relying on hidden chat history or blind full-repo scans.
• Read-only and write profiles
  Separates safe context loading from explicit memory writes, helping teams control what agents may update.

ITHZ MCP fordeler

• Complements Git
  Git remains the source-code history, while ITHZ MCP adds a separate layer for project memory.
• Improves auditability
  Decisions, gates, risks, claims, and checkpoints become inspectable records.
• Reduces context drift
  Agents can start from durable project memory instead of forgetting prior decisions between sessions.
• Helps large repositories
  Generated files and noisy outputs can be pruned from memory intake to keep context useful.
• Supports claim discipline
  Allowed and blocked claims help agents avoid repeating statements that are not approved.
• Keeps memory project-owned
  Important context stays with the project rather than being hidden inside vendor chat history.
• Speeds up agent onboarding
  New agent sessions can load bounded context packs before reading broad source trees.
• Improves human-agent handoff
  Developers can see what an agent did, why it did it, and what should happen next.
• Supports safer local workflows
  Memory can remain local-first and explicit instead of becoming uncontrolled cloud sync.
• Creates clearer project continuity
  The project keeps a durable trail of decisions, evidence, risks, and next steps across AI-assisted work.

Hvem bør bruke ITHZ MCP?

• Research teams
  Teams working with evolving evidence, hypotheses, publication boundaries, and claim discipline.
• AI workflow leads
  People designing how agents should read context, write memory, checkpoint work, and hand off safely.
• Engineering teams
  Teams that need shared decisions, risks, gates, and workflow context to survive across agent sessions.
• Platform engineers
  Engineers responsible for repeatable local-first developer workflows and safer AI tooling adoption.
• Technical founders
  Solo builders or small-team founders who want durable project memory without introducing a cloud knowledge base.
• Project maintainers
  Maintainers who want future AI runs to start from the actual project state instead of reconstructing everything.
• Technical reviewers
  Reviewers who need to inspect what was decided, what was verified, and which claims are allowed or blocked.
• Large-repo developers
  Developers working in repositories where repeated full scans are noisy, slow, or too expensive.
• AI-assisted developers
  Developers who use Codex, Claude Code, Cursor, or similar tools and need agents to remember project context safely.
• Organizations evaluating agents
  Companies testing agentic development workflows while needing auditability and control.

---

Merker

---

ITHZ MCP-alternativer

---

Teknologier brukt

Vis alle teknologier