Update for skill.ts

[drQedwards]

Card entry for pmll is now in the appropriate directory.

{
title: "PMLL",
description: "Gives AI agents persistent spatial compute memory so they can retain long-term context, form symbiotic memory layers, and maintain durable state across sessions; supports PPM-based context stitching, Context+ pipelines, and supermodeltools/cli for graphing/analysis. On-chain commitment anchoring (32-byte hashes via Soroban) is planned.",
pathLabel: "drQedwards/pmll",
copyValue: "https://github.com/drQedwards/pmll/blob/main/SKILL.md",
},

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Adds a new ecosystem/skill card entry (PMLL) to the site’s skills data so it can be surfaced in the UI.

Changes:

• Introduces a new ECOSYSTEM_CARDS entry for “PMLL” with description and links.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[kaankacar]

🤖 Automated message from Kaan's Automated Triage Bot.

👀 Picked this up — a review will follow shortly.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 01b87d2329

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Move PMLL card out of the Contextio object

The inserted { starts the PMLL entry before the Contextio SDK object is closed, which makes site/src/data/skills.ts invalid TypeScript when the site parses this data module. This prevents the skills page build from compiling; close the Contextio entry first and add PMLL as its own element in ECOSYSTEM_CARDS.

Useful? React with 👍 / 👎.

[kaankacar]

🤖 Automated message from Kaan's Automated Triage Bot.

Thanks for this — persistent memory for agents is a real gap, and I appreciate that both the card and SKILL.md say plainly that the Soroban anchoring is planned rather than dressing it up as shipped. That honesty made this much easier to review. For what it's worth, the checkable parts hold up: the repo is public and MIT, pmll-memory-mcp really is published on both PyPI (2.0.0) and npm (2.0.2) under your name, contextplus / supermodeltools / PPM all resolve, the testnet passphrase in stellar.toml matches SEP-1, and your card's fields line up with EcosystemCardSource.

Three things before it can go on the catalog:

1. The entry is nested inside the Contextio object, so skills.ts isn't valid TypeScript as it stands. Copilot and Codex both caught this and they're right — the Contextio entry needs closing first:

    copyValue:
      "https://github.com/Eras256/Contextio/blob/main/packages/sdk/SKILL.md",
  },
  {
    title: "PMLL",
    // ...
  },
] as const;

2. SKILL.md looks truncated. The file at your copyValue URL is 1122 bytes and stops partway through Quick start step 1 — the ```bash fence opened after pip install pmll-memory-mcp is never closed, and there are no further steps. Separately, every other skill in the catalog uses name: in its frontmatter (the Agent Skills field); yours has title:, which agent loaders won't pick up.

3. The Stellar side is the part I can't settle on my own. Searching the repo, the only two files that mention Stellar or Soroban at all are SKILL.md and stellar.toml — and the latter is still a placeholder (pmll_anchor = "REPLACE_WITH_PMll_ANCHOR_CONTRACT_ID"), with no Rust or Soroban contract anywhere in the tree. Every entry currently on the catalog ships working Stellar integration, so whether a listing can go up on a planned one is a maintainer judgment rather than mine. I've flagged it for @kaankacar and @ElliotFriend to weigh in.

I held off pushing the brace fix myself, since you'll need another push for the SKILL.md content anyway and it's easiest to do both at once. Happy to re-review as soon as you do.

[kaankacar]

🤖 Automated message from Kaan's Automated Triage Bot.

Thanks for the quick turnaround — closing the Contextio object was exactly the right move, and your card's four fields match EcosystemCardSource exactly. Closing it did leave the older }, further down with nothing left to close, so the file still didn't parse. That's a one-character thing and not worth sending you round again, so I've pushed the fix to your branch (a7d8c53): stray brace gone, indentation lined up with the neighbouring entries, your wording untouched. The diff is now a single clean entry.

Two things from last time still need a push to drQedwards/pmll, which hasn't changed since Aug 15:

• SKILL.md is still the same 1122 bytes and still stops partway through Quick start step 1 — the bash code fence opened after pip install pmll-memory-mcp is never closed, and there's no step 2.
• The frontmatter still reads title:. Agent Skills loaders key off name:, so an agent won't pick the skill up as written.

The Stellar question is unchanged, and it genuinely isn't mine to settle: there's still no Soroban contract anywhere in the tree (no .rs, no Cargo.toml) and stellar.toml is still REPLACE_WITH_PMll_ANCHOR_CONTRACT_ID, so the anchoring remains what you say it is — planned. Whether the catalog lists on that basis is @kaankacar and @ElliotFriend's call, and I've added today's findings to the open hold rather than merging. Nothing I checked looks like a security concern.

[drQedwards]

🤖 Automated message from Kaan's Automated Triage Bot.

Thanks for the quick turnaround — closing the Contextio object was exactly the right move, and your card's four fields match EcosystemCardSource exactly. Closing it did leave the older }, further down with nothing left to close, so the file still didn't parse. That's a one-character thing and not worth sending you round again, so I've pushed the fix to your branch (a7d8c53): stray brace gone, indentation lined up with the neighbouring entries, your wording untouched. The diff is now a single clean entry.

Two things from last time still need a push to drQedwards/pmll, which hasn't changed since Aug 15:

• SKILL.md is still the same 1122 bytes and still stops partway through Quick start step 1 — the bash code fence opened after pip install pmll-memory-mcp is never closed, and there's no step 2.
• The frontmatter still reads title:. Agent Skills loaders key off name:, so an agent won't pick the skill up as written.

The Stellar question is unchanged, and it genuinely isn't mine to settle: there's still no Soroban contract anywhere in the tree (no .rs, no Cargo.toml) and stellar.toml is still REPLACE_WITH_PMll_ANCHOR_CONTRACT_ID, so the anchoring remains what you say it is — planned. Whether the catalog lists on that basis is @kaankacar and @ElliotFriend's call, and I've added today's findings to the open hold rather than merging. Nothing I checked looks like a security concern.

@kaankacar issue one resolved. Quick check before I move to resolve the front end matter.

[kaankacar]

🤖 Automated message from Kaan's Automated Triage Bot.

That did it — both of my mechanical points are closed. SKILL.md (ffe4827) is complete now: the bash fence closes, steps 2 and 3 are there, and the frontmatter reads name: pmll. I re-checked the rest while I was in there, and it all holds up — the card URL and repo both return 200, pmll-memory-mcp is live on PyPI and npm at 2.0.2 with a pmll-memory-mcp entry point on both sides so the pip and npx paths in Quick start work exactly as written, and every tool you name in step 3 (init, peek, set, resolve, flush, the graph ops) is in your MCP-tools.json. The testnet passphrase in stellar.toml is the correct SEP-1 value per the docs, and I found nothing that looks like a security concern.

So the only thing left is the one that isn't mine to settle. There's still no Soroban contract in the tree — no .rs, no Cargo.toml, and stellar.toml is still REPLACE_WITH_PMll_ANCHOR_CONTRACT_ID — so the anchoring is precisely what you say it is: planned. Every entry on the catalog today ships working Stellar integration, so whether a listing can go up ahead of that is @kaankacar and @ElliotFriend's call. I've added today's findings to their open hold. Nothing more needed from you while it sits there, and I'll merge as soon as they release it.

[kaankacar]

🤖 Automated message from Kaan's Automated Triage Bot.

Thanks for sticking with this — and pmll-anchor is a real step forward. I read it end to end: store writes a 32-byte commitment behind admin.require_auth(), extends the TTL and emits an event, which is exactly the design your SKILL.md describes; the extend_ttl(&key, threshold, extend_to) shape matches the storage-strategies docs, target/wasm32v1-none/release/ in DEPLOY.md is the current build path, and soroban-sdk 27.0.6 is live on crates.io. Nothing in it looked like a security concern.

Two things in the latest push need one more commit, though:

1. The merge at fdde01e re-broke the PMLL card — it now carries every field twice inside a single object:

    copyValue:
      "https://github.com/drQedwards/pmll/blob/main/SKILL.md",
    title: "PMLL",          // <- a second copy starts here
    description: "...",
    pathLabel: "drQedwards/pmll",
    copyValue: "https://github.com/drQedwards/pmll/blob/main/SKILL.md",
  },

Duplicate properties in an object literal are a TypeScript error (TS1117), so pnpm lint:ts won't pass. More importantly the two copies don't say the same thing: the second drops the "anchoring is planned" wording and describes the memory as "on Stellar" outright. Since the later key wins, that second one is what would actually render — and it contradicts your own SKILL.md, which still says off-chain today. Please keep just one block; I'd suggest the first, which is the wording in your PR description. I deliberately didn't pick for you this time — last round it was a stray brace, but this is your wording, not something I should straighten on your behalf.

2. The new ppm card points at a stale copy of this same skill. drQedwards/ppm/SKILL.md is 1122 bytes and byte-for-byte the version of pmll/SKILL.md from before you fixed it: frontmatter still title: rather than the Agent Skills name:, and it still stops partway through Quick start step 1 with the ```bash fence never closed. Its H1 is # PMLL, so it reads as the PMLL skill rather than a PPM one, and the only Stellar-mentioning file in that repo is the same placeholder stellar.toml. If PPM is meant to be its own listing it needs its own finished SKILL.md; if it isn't, dropping that entry is the simplest fix.

Copilot's and Codex's earlier note about the entry nesting is resolved and I'm not reopening it. The catalog-bar question is unchanged and still isn't mine to settle: stellar.toml is still REPLACE_WITH_PMll_ANCHOR_CONTRACT_ID, so the anchor is written but not deployed, and your SKILL.md accurately still calls it planned. That sits with @kaankacar and @ElliotFriend, and I've added today's findings — the contract included — to their open hold. Happy to re-review as soon as you push.