- evidence[0]: unknown sourceId 'bjjf-leg-attacks-1' (not in pipeline/sources/)
- evidence[1]: unknown sourceId 'bjj-world-ashi-garami-made-simple-zahabi-bundle-review' (not in pipeline/sources/)
- evidence[2]: unknown sourceId 'bjjf-outside-ashi-reloaded-by-jason-rau' (not in pipeline/sources/)
Closes half of g-outside-ashi-garami-no_threatens (leaf position with no submission threats recorded). Targets the existing outside-heel-hook node: the BJJ Fanatics blog names the heel hook as the signature attack from outside ashi garami, and the BJJ World review of the Zahabi outside/cross ashi material specifies 'outside heel hooks' as the finishing mechanic taught — no ledger source ties this position to the inside heel hook, so no edge to inside-heel-hook is proposed. Threat level 'high' follows the blog's language: the position 'provides a strong level of control' and 'offers a great place from which to launch a heel hook attack', in explicit contrast to the straight ankle lock which the same article calls 'less likely' from here; the heel hook is the one attack every consulted source foregrounds for this position. Snapshot fidelity caveat: ledger snapshots for this run were reconstructed from search extracts (egress-blocked environment) — re-verify against live pages before or after applying.
Evidence
bjjf-leg-attacks-1article body, outside ashi garami section
“This position provides a strong level of control over the legs of the opponent and offers a great place from which to launch a heel hook attack.”
bjj-world-ashi-garami-made-simple-zahabi-bundle-reviewreview body, finishes overview
“Firas first talks finishes, explaining the mechanics of outside heel hooks, straight ankle locks, and Aoki locks.”
bjjf-outside-ashi-reloaded-by-jason-rauproduct description, first paragraph
“seamlessly combining two of the most effective leg submissions in Brazilian Jiu-Jitsu: the ankle lock and the heel hook”
Payload
{
"edge": {
"id": "e-outside-ashi-ohh",
"type": "THREATENS",
"source": "outside-ashi-garami",
"target": "outside-heel-hook",
"threatLevel": "high"
}
}node tools/apply-proposal.mjs p-r20260716a-outside-ashi-threatens-ohh [--reject] [--note "…"]
- evidence[0]: unknown sourceId 'bjjf-outside-ashi-reloaded-by-jason-rau' (not in pipeline/sources/)
- evidence[1]: unknown sourceId 'bjjf-leg-attacks-1' (not in pipeline/sources/)
Closes half of g-outside-ashi-garami-no_threatens (leaf position with no submission threats recorded). Depends on the sibling add_node p-r20260716a-straight-ankle-lock-node introducing the target id. Threat level 'medium' is a deliberate middle reading of the two sources: Jason Rau's dedicated outside-ashi instructional presents the ankle lock as one of the two headline attacks chained from this position ('the ankle lock and the heel hook'), which rules out 'low' as an understatement of how the position is actually taught; but the BJJ Fanatics blog explicitly hedges the same attack from the double-outside configuration ('less likely here because of the placement of the legs, unless the position of the feet are adjusted'), which rules out 'high'. Both quotes are attached so the reviewer sees the tension rather than a blended claim. Snapshot fidelity caveat: ledger snapshots for this run were reconstructed from search extracts (egress-blocked environment) — re-verify against live pages before or after applying.
Evidence
bjjf-outside-ashi-reloaded-by-jason-rauproduct description, first paragraph
“This comprehensive course unlocks the hidden potential of outside ashi garami by seamlessly combining two of the most effective leg submissions in Brazilian Jiu-Jitsu: the ankle lock and the heel hook.”
bjjf-leg-attacks-1article body, outside ashi garami section
“The straight ankle lock submission is less likely here because of the placement of the legs, unless the position of the feet are adjusted to place them on the hips”
Payload
{
"edge": {
"id": "e-outside-ashi-sal",
"type": "THREATENS",
"source": "outside-ashi-garami",
"target": "straight-ankle-lock",
"threatLevel": "medium"
}
}node tools/apply-proposal.mjs p-r20260716a-outside-ashi-threatens-sal [--reject] [--note "…"]
- evidence[0]: unknown sourceId 'bjjf-leg-attacks-1' (not in pipeline/sources/)
- evidence[1]: unknown sourceId 'bjjf-outside-ashi-reloaded-by-jason-rau' (not in pipeline/sources/)
- evidence[2]: unknown sourceId 'bjj-world-ashi-garami-made-simple-zahabi-bundle-review' (not in pipeline/sources/)
- nodes/straight-ankle-lock.json.sources[1]: bad source kind 'instructional-toc'
Prerequisite for closing g-outside-ashi-garami-no_threatens: the sourced threat set for outside-ashi-garami includes the straight ankle lock, but no straight-ankle-lock node exists in graph/data/nodes/. This adds the submission node so a sibling add_edge proposal (p-r20260716a-outside-ashi-threatens-sal) can target it — the gate is batch-aware. Shape mirrors the existing submission nodes (inside-heel-hook, outside-heel-hook); no threatInputs are included because none of the ledger snapshots describe the grip/exposure mechanics of the submission itself, and inventing them would exceed the evidence. Alias 'ankle lock' appears verbatim in the Rau description. Initial provenance.status is 'settled' because three independent tier-1 sources (a BJJF blog article, a BJJF product page, and a BJJ World analyst review) all name the technique consistently — this is an initial status on a new node, not a status transition on an existing one. Snapshot fidelity caveat: ledger snapshots for this run were reconstructed from search extracts (egress-blocked environment) — re-verify against live pages before or after applying.
Evidence
bjjf-leg-attacks-1article body, opening paragraphs
“The ashi garami position is the primary controlling position from which to attack the straight ankle lock.”
bjjf-outside-ashi-reloaded-by-jason-rauproduct description, first paragraph
“seamlessly combining two of the most effective leg submissions in Brazilian Jiu-Jitsu: the ankle lock and the heel hook”
bjj-world-ashi-garami-made-simple-zahabi-bundle-reviewreview body, finishes overview
“explaining the mechanics of outside heel hooks, straight ankle locks, and Aoki locks”
Payload
{
"node": {
"id": "straight-ankle-lock",
"kind": "submission",
"name": "Straight ankle lock",
"aliases": [
"ankle lock"
],
"provenance": {
"status": "settled",
"sources": [
{
"label": "The Powerful Ashi Garami & Outside Ashi Garami — BJJ Fanatics blog",
"kind": "article",
"url": "https://bjjfanatics.com/blogs/news/leg-attacks-1"
},
{
"label": "Outside Ashi Reloaded by Jason Rau — BJJ Fanatics",
"kind": "instructional-toc",
"url": "https://bjjfanatics.com/products/outside-ashi-reloaded-by-jason-rau"
},
{
"label": "Ashi Garami Made Simple Firas Zahabi DVD Bundle Review — BJJ World",
"kind": "article",
"url": "https://bjj-world.com/ashi-garami-made-simple-firas-zahabi-dvd-bundle-review/"
}
]
}
}
}node tools/apply-proposal.mjs p-r20260716a-straight-ankle-lock-node [--reject] [--note "…"]
Grounds a mapping the pipeline already relies on but nothing evidences. aliases is a structure field (framework 6.3: names and grips are not identity), so this changes no claim, status or edge — it records that a tier-1 source calls the inside heel hook "the inverted". The value is the existing array plus one entry. This matters beyond the node page: youtube/lexicon.json routes "inverted heel hook" to inside-heel-hook when tagging every ledger entry, so the term is load-bearing for tagging precision while resting on no quoted source. This proposal supplies the quote or forces the lexicon entry to be reconsidered.
Risks the agent flagged
- Ryan never says "the inverted heel hook is the inside heel hook". The identification is inferred: he names exactly two mechanics at 416.46s (inside and outside), then for the rest of the video uses "inverted" only ever against "outside", never alongside "inside". That is an inference from contrast, not a stated equation.
- Whisper renders the term as "inverted heel look" at 527.98s and "the outside heel look" at 529.48s, so the alias string itself should be confirmed by ear before it is written into the graph.
- This is one lineage's idiolect. "Inverted heel hook" is widespread in no-gi, but this proposal evidences only that Gordon Ryan uses it for the finish he contrasts with the outside heel hook — it does not establish consensus, and the reviewer should not read the alias as settled terminology on one source.
- update_field replaces the whole aliases array. The value repeats the node's existing "medial heel hook" so nothing is dropped, but that must be checked against graph/data/nodes/inside-heel-hook.json at apply time rather than assumed.
- The tagging lexicon already asserts this mapping (youtube/lexicon.json: inside-heel-hook extra "inverted heel hook", "reverse heel hook"), so every moment tagged from that term across the ledger currently depends on an assertion no node evidences. Approving this makes the lexicon override redundant; rejecting it means the lexicon entry is the thing that needs revisiting, not this proposal.
Evidence— machine transcription (Whisper) — near-verbatim, but not the speaker’s own subtitles
yt-_IxtobCktAQ6:56establishes that there are exactly two heel hooks in play — inside and outside
All right. Now, when you go to inside-ashi, that requires you to have a strong outside heel hook breaking mechanics, uh, breaking mechanics, which most people don't have, which is why most people don't use it. But it's very important to have strong inside heel hook mechanics and strong outside heel hook mechanics so you can play these two positions off one another. If my opponent is turning towards the camera, it's the easiest thing in the world to put a heel hook in with an inverted heel hook. But if Penardo goes to roll away from me, we just go in, we control his toes, and we take him to the center of our chest. So when he goes to make that roll through, he can't now, or else he'll break his own leg.
yt-_IxtobCktAQ6:20first use of "inverted heel hook" for the finish he is teaching
It's definitely important, but it's, it's the smallest part of the equation. Well, but can you show any concept here about the heel hook that you think would be interesting for everybody to see just so they have an idea of what's coming in? Yeah. So, um, going off of this, uh, this heel hook that we hit at ADCC with, uh, with Nicky Rod, for example, where when I go for an inverted heel hook and he wants to roll in this direction, everyone knows that. Okay. Now you're in cross-ashigurami, my opponent wants to roll in this direction. And if he rolls in this direction, you can either keep rolling and roll out, or you can take your back like Felipe penned it to me two times in a row. Okay.
yt-_IxtobCktAQ8:49"the inverted" used as the counterpart term to "the outside"
So the idea here is, we go from... We go from cross-ashi. We want to hit an inverted heel look. Because ultimately, the inverted is mechanically stronger than the outside heel look. But everyone knows, you know, people have seen this through leg locks now. They know, man, he goes cross-ashi, roll, roll, roll. So as he goes to roll, we go in, we just grab the toes, and we take his leg's center chest.
Contradictions
yt-_IxtobCktAQ6:48
⚠ This alias rests on "inside" and "outside" being stable labels for the two heel hooks in this video, but eight seconds earlier the same speaker attaches "inside" to the position and "outside" to the heel hook attacked from it. The adjectives are doing two different jobs in one passage, which is the main reason to distrust reading "inverted" off as "inside" — it is recorded here rather than buried so the reviewer can reject the alias on exactly this ground.
So I wanted to avoid this. The best way to avoid this is by going into inside-ashi. All right. Now, when you go to inside-ashi, that requires you to have a strong outside heel hook breaking mechanics, uh, breaking mechanics, which most people don't have, which is why most people don't use it. But it's very important to have strong inside heel hook mechanics and strong outside heel hook mechanics so you can play these two positions off one another. If my opponent is turning towards the camera, it's the easiest thing in the world to put a heel hook in with an inverted heel hook. But if Penardo goes to roll away from me, we just go in, we control his toes, and we take him to the center of our chest.
Payload
{
"field": "aliases",
"value": [
"medial heel hook",
"inverted heel hook"
]
}node tools/apply-proposal.mjs p-r20260820a-inside-heel-hook-aliases-inverted [--reject] [--note "…"]
Closes g-inside-heel-hook-unsourced (R2). The node currently cites only "Standard grappling terminology" with no URL, so nothing on it is traceable. This entry is the strongest ledger material for the node: a tier-1 source that names the inside heel hook explicitly (most heel-hook videos in the ledger, including both Lachlan Giles entries, say only "heel hook"), treats it as a distinct set of mechanics from the outside heel hook, and ranks the two against each other. Only the source is added — no claim, no edge — because the same passage uses inside/outside inconsistently across positions and finishes (see contradictions).
Risks the agent flagged
- Whisper ASR, and demonstrably imperfect right where it matters: the same passage renders "heel hook" as "heel look" at 527.98s and again in the evidence quote at 529.48s. The quote is verbatim to the stored transcript, not necessarily to Ryan's mouth.
- "The inverted is mechanically stronger than the outside" is one elite competitor's opinion delivered while demonstrating on a talk-show partner, not a measured or consensus claim. It should be recorded as Ryan's framing if it is ever promoted into a claim rather than a source.
- The video uses "inside" and "outside" of positions and of heel hooks in adjacent sentences with opposite pairings (see contradictions), so any downstream reading of this source needs the surrounding 30 seconds, not the quote alone.
- This closes g-inside-heel-hook-unsourced only. The node stays 'settled' on one URL-bearing source, so g-inside-heel-hook-insufficient_sources remains open and needs a source independent of BJJ Fanatics.
Evidence— machine transcription (Whisper) — near-verbatim, but not the speaker’s own subtitles
yt-_IxtobCktAQ6:56ADCC Nicky Rod sequence — names the two heel hooks as the paired mechanics
All right. Now, when you go to inside-ashi, that requires you to have a strong outside heel hook breaking mechanics, uh, breaking mechanics, which most people don't have, which is why most people don't use it. But it's very important to have strong inside heel hook mechanics and strong outside heel hook mechanics so you can play these two positions off one another. If my opponent is turning towards the camera, it's the easiest thing in the world to put a heel hook in with an inverted heel hook. But if Penardo goes to roll away from me, we just go in, we control his toes, and we take him to the center of our chest. So when he goes to make that roll through, he can't now, or else he'll break his own leg.
yt-_IxtobCktAQ8:49second pass over the same sequence, comparing the two finishes
So the idea here is, we go from... We go from cross-ashi. We want to hit an inverted heel look. Because ultimately, the inverted is mechanically stronger than the outside heel look. But everyone knows, you know, people have seen this through leg locks now. They know, man, he goes cross-ashi, roll, roll, roll. So as he goes to roll, we go in, we just grab the toes, and we take his leg's center chest.
Contradictions
yt-_IxtobCktAQ6:48
⚠ Eight seconds before the evidence quote, Ryan pairs the *inside* position with the *outside* heel hook. The graph pairs them the other way: saddle (inside/inside signature) THREATENS inside-heel-hook (e-saddle-ihh). Either "inside ashi" names a structure that is not the saddle, or the inside/outside adjective does not carry across from the position to the finish in this lineage's speech. A reviewer must not read this source's "inside" as automatically the graph's inside-heel-hook, which is why this proposal adds only a source and proposes no THREATENS edge from it.
So I wanted to avoid this. The best way to avoid this is by going into inside-ashi. All right. Now, when you go to inside-ashi, that requires you to have a strong outside heel hook breaking mechanics, uh, breaking mechanics, which most people don't have, which is why most people don't use it. But it's very important to have strong inside heel hook mechanics and strong outside heel hook mechanics so you can play these two positions off one another. If my opponent is turning towards the camera, it's the easiest thing in the world to put a heel hook in with an inverted heel hook. But if Penardo goes to roll away from me, we just go in, we control his toes, and we take him to the center of our chest.
Payload
{
"source": {
"label": "Gordon Ryan, \"How To Perfect Leglocks In Jiu Jitsu No Gi\" — Bernardo Faria BJJ Fanatics (YouTube)",
"url": "https://www.youtube.com/watch?v=_IxtobCktAQ",
"kind": "caption"
}
}node tools/apply-proposal.mjs p-r20260820a-inside-heel-hook-source-ryan [--reject] [--note "…"]
Closes g-outside-heel-hook-unsourced (R2). The node currently cites only the unlinkable label "Standard grappling terminology". This entry names the outside heel hook six times and, unusually for the ledger, treats it as a position-specific finish rather than a generic "heel hook": the defender's roll exposes the heel to it (448.98s) and there is a named position both players attack it from (718.84s). Source only — no THREATENS edge is proposed, because the position Ryan attaches it to ("mutual ashi") is not in the graph and is never described structurally.
Risks the agent flagged
- Whisper ASR, and this transcript mangles the term nearby — "outside heel look" at 529.48s — so the reviewer should confirm the exact wording of both evidence quotes by ear before trusting them as Ryan's phrasing.
- The contradiction rests on reading an analogy ("the 50-50 of outside heel hooks"), not a definition. Ryan never says 50-50 does or does not threaten the outside heel hook, and a reviewer could reasonably read the line the opposite way.
- "Mutual ashi" has no node in the graph and is given no structural description here beyond both players having access to outside heel hooks, so this source cannot be used to add or place that position — only to record that Ryan distinguishes it.
- This closes g-outside-heel-hook-unsourced only. The node stays 'settled' on one URL-bearing source, leaving g-outside-heel-hook-insufficient_sources open; the ledger's Zahabi/Rau text sources are the obvious independent second.
Evidence— machine transcription (Whisper) — near-verbatim, but not the speaker’s own subtitles
yt-_IxtobCktAQ7:28ADCC sequence — the outside heel hook as the finish the defender rolls into
Now he wants to continue rolling in this direction. And as he does so, I just make a pass from cross-ashigurami to inside-ashigurami. So now when he goes to start rolling in this direction, he exposes his own heel to an outside heel hook. Okay? And now, if he continues rolling in this direction, I just roll through with him and I break him. What Nicky Rod did was actually smart.
yt-_IxtobCktAQ11:58training-methods section, naming the position both players attack the outside heel hook from
If I leg lock him, we switch positions. So we do positional rounds where we start in 50-50, cross ashi. We can do positional rounds where we start in mutual ashi, like so, where we both have access to outside heel hooks. This is basically the 50-50 of outside heel hooks. And we just do these positions where you start in these positions, and you spend hours and hours and hours doing live rolls from these positions.
Contradictions
yt-_IxtobCktAQ11:58
⚠ The graph records fifty-fifty THREATENS outside-heel-hook at threatLevel 'high' (edges/e-5050-ohh.json). Ryan instead names a different, unmodelled position — "mutual ashi" — as the place both players attack the outside heel hook, and uses 50-50 only as the analogy for it. "The 50-50 of outside heel hooks" reads as: 50-50 is that kind of mutual position for some *other* finish, i.e. 50-50 is not itself the outside-heel-hook position in his usage. This is an argument against the existing edge, not against adding this source; it is recorded so the reviewer sees the disagreement before the node accumulates more outside-heel-hook material.
If I leg lock him, we switch positions. So we do positional rounds where we start in 50-50, cross ashi. We can do positional rounds where we start in mutual ashi, like so, where we both have access to outside heel hooks. This is basically the 50-50 of outside heel hooks. And we just do these positions where you start in these positions, and you spend hours and hours and hours doing live rolls from these positions. If you're in a, you know, a six-minute match, and you're rolling with your buddy, 90% of the roll is going to be an open guard, top or bottom.
Payload
{
"source": {
"label": "Gordon Ryan, \"How To Perfect Leglocks In Jiu Jitsu No Gi\" — Bernardo Faria BJJ Fanatics (YouTube)",
"url": "https://www.youtube.com/watch?v=_IxtobCktAQ",
"kind": "caption"
}
}node tools/apply-proposal.mjs p-r20260820a-outside-heel-hook-source-ryan [--reject] [--note "…"]
Closes g-saddle-unsourced (R2: "no URL-bearing source — every node needs at least one citable source"). The saddle node currently cites only an unlinkable label ("Modern leglock canon (inside sankaku)"). This is the cleanest saddle definition in the ledger: a tier-1 instructor names the position twice in the first 40 seconds and describes it as state rather than mechanism (framework 6.2) — the leg passed across the attacker's legs, a reap, both feet across the body — which is exactly the shape of definition the node needs. No claim about the position's threats or signature is proposed; only the citable source is added.
Risks the agent flagged
- The transcript is Whisper ASR, not the creator's own subtitles. The word carrying the structural content of the first quote — "reap" — is a single short ASR token, and the reviewer should hear 10.0s before treating it as Giles's word.
- Giles is demonstrating on a partner, so the quote describes what the camera shows ("I've come across here", "both my feet are across the body here"). The per-leg inside/outside signature the graph keys identity to (framework 6.3) is never stated in words; this source names and locates the position, it does not evidence the signature already recorded on the node.
- Giles calls the position "the saddle" unqualified here, but in his other ledger entry (yt-w-W0ug7Edag at 36.5s) the same instructor says it "got a bunch of different names" and picks "saddle" "for the sake of giving it a name". This source is therefore weak evidence about terminology and strong evidence only about the configuration being taught.
- This closes g-saddle-unsourced but not g-saddle-insufficient_sources: the node stays 'settled' on one URL-bearing source, and the second must be independent of Absolute MMA St Kilda.
Evidence— machine transcription (Whisper) — near-verbatim, but not the speaker’s own subtitles
yt-oeyvy8jJIbA0:10opening, naming and describing the position before any attack
Today's class is an introduction to the heel hook from the saddle, so this is the most stable position I think you can get to start to learn how to do the heel hook. We'll start from the saddle, which is this position where I've got, basically I've passed their leg across my legs and I've got a reap, so I've come across here, both my feet are across the body here. Okay, this is the saddle, alright, so when we're in the saddle, we're going to just be trying to keep, I want to basically, mostly be trying to stop my opponent from turning away. Okay, I'm going to be looking for the heel hook, alright, I'm going to try and capture this heel and he's going to be trying to turn away usually to hide that, make that difficult to get, okay. So, a couple of things that stop our opponent from turning away, okay.
yt-oeyvy8jJIbA0:25second naming of the position, control priority
Today's class is an introduction to the heel hook from the saddle, so this is the most stable position I think you can get to start to learn how to do the heel hook. We'll start from the saddle, which is this position where I've got, basically I've passed their leg across my legs and I've got a reap, so I've come across here, both my feet are across the body here. Okay, this is the saddle, alright, so when we're in the saddle, we're going to just be trying to keep, I want to basically, mostly be trying to stop my opponent from turning away. Okay, I'm going to be looking for the heel hook, alright, I'm going to try and capture this heel and he's going to be trying to turn away usually to hide that, make that difficult to get, okay. So, a couple of things that stop our opponent from turning away, okay. The main one, if you can get it, is getting this second leg here, okay.
Contradictions
yt-_IxtobCktAQ7:23
⚠ The tagging lexicon maps "cross ashi" / "cross ashi garami" onto this node (pipeline/youtube/lexicon.json, saddle.extra), but Gordon Ryan treats cross-ashigurami and inside-ashigurami as two distinct positions he passes *between*, and at 525.4-529.5s attacks the inverted heel hook from cross-ashi while at 408.3s pairing inside-ashi with the outside heel hook. If "cross ashi" is not the saddle in this lineage's usage, the three saddle-tagged moments in yt-_IxtobCktAQ are mis-tagged and the lexicon alias is wrong. Nothing in the Giles entry cited as evidence here depends on that alias — Giles says "saddle" outright — but the reviewer should settle the alias before treating the two videos as covering the same node.
So when he goes to make that roll through, he can't now, or else he'll break his own leg. Now he wants to continue rolling in this direction. And as he does so, I just make a pass from cross-ashigurami to inside-ashigurami. So now when he goes to start rolling in this direction, he exposes his own heel to an outside heel hook. Okay? And now, if he continues rolling in this direction, I just roll through with him and I break him.
Payload
{
"source": {
"label": "Lachlan Giles, \"Learn Heel Hooks\" — Absolute MMA St Kilda (YouTube)",
"url": "https://www.youtube.com/watch?v=oeyvy8jJIbA",
"kind": "caption"
}
}node tools/apply-proposal.mjs p-r20260820a-saddle-source-giles [--reject] [--note "…"]